Datasets
Models
Downloads: 1
7310 lines (7310 with data), 830.7 kB
{
"cells": [
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"id": "VGCYrVKDW2k6"
},
"outputs": [],
"source": [
"# For tips on running notebooks in Google Colab, see\n",
"# https://pytorch.org/tutorials/beginner/colab\n",
"%matplotlib inline\n",
"# from google.colab import drive\n",
"# drive.mount('/content/gdrive/', force_remount=True)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "RiiUBIxUW2k9"
},
"source": [
"\n",
"# Transfer Learning for Computer Vision Tutorial\n",
"**Author**: [Sasank Chilamkurthy](https://chsasank.github.io)\n",
"\n",
"In this tutorial, you will learn how to train a convolutional neural network for\n",
"image classification using transfer learning. You can read more about the transfer\n",
"learning at [cs231n notes](https://cs231n.github.io/transfer-learning/)_\n",
"\n",
"Quoting these notes,\n",
"\n",
" In practice, very few people train an entire Convolutional Network\n",
" from scratch (with random initialization), because it is relatively\n",
" rare to have a dataset of sufficient size. Instead, it is common to\n",
" pretrain a ConvNet on a very large dataset (e.g. ImageNet, which\n",
" contains 1.2 million images with 1000 categories), and then use the\n",
" ConvNet either as an initialization or a fixed feature extractor for\n",
" the task of interest.\n",
"\n",
"These two major transfer learning scenarios look as follows:\n",
"\n",
"- **Finetuning the ConvNet**: Instead of random initialization, we\n",
" initialize the network with a pretrained network, like the one that is\n",
" trained on imagenet 1000 dataset. Rest of the training looks as\n",
" usual.\n",
"- **ConvNet as fixed feature extractor**: Here, we will freeze the weights\n",
" for all of the network except that of the final fully connected\n",
" layer. This last fully connected layer is replaced with a new one\n",
" with random weights and only this layer is trained.\n"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"id": "XyEZTo0NW2k_",
"outputId": "b2b6c910-fda6-4006-9c16-225acbe06ce2"
},
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Time in seconds since beginning of run: 1685206884.1100936\n",
"Sat May 27 17:01:24 2023\n"
]
}
],
"source": [
"# License: BSD\n",
"# Author: Sasank Chilamkurthy\n",
"\n",
"from __future__ import print_function, division\n",
"\n",
"import torch\n",
"import torch.nn as nn\n",
"import torch.optim as optim\n",
"from torch.optim import lr_scheduler\n",
"import torch.backends.cudnn as cudnn\n",
"import numpy as np\n",
"import torchvision\n",
"from torchvision import datasets, models, transforms\n",
"import matplotlib.pyplot as plt\n",
"import time\n",
"import os\n",
"import copy\n",
"\n",
"cudnn.benchmark = True\n",
"plt.ion() # interactive mode\n",
"\n",
"seconds = time.time()\n",
"print(\"Time in seconds since beginning of run:\", seconds)\n",
"local_time = time.ctime(seconds)\n",
"print(local_time)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "iehYC9-cW2lA"
},
"source": [
"## Load Data\n",
"\n",
"We will use torchvision and torch.utils.data packages for loading the\n",
"data.\n",
"\n",
"The problem we're going to solve today is to train a model to classify\n",
"**ants** and **bees**. We have about 120 training images each for ants and bees.\n",
"There are 75 validation images for each class. Usually, this is a very\n",
"small dataset to generalize upon, if trained from scratch. Since we\n",
"are using transfer learning, we should be able to generalize reasonably\n",
"well.\n",
"\n",
"This dataset is a very small subset of imagenet.\n",
"\n",
".. Note ::\n",
" Download the data from\n",
" [here](https://download.pytorch.org/tutorial/hymenoptera_data.zip)\n",
" and extract it to the current directory.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"id": "7UvqjeCdW2lA"
},
"outputs": [],
"source": [
"# Data augmentation and normalization for training\n",
"# Just normalization for validation\n",
"data_transforms = {\n",
" 'train': transforms.Compose([\n",
" transforms.RandomResizedCrop(224),\n",
" transforms.RandomHorizontalFlip(),\n",
" transforms.ToTensor(),\n",
" transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
" ]),\n",
" 'val': transforms.Compose([\n",
" transforms.Resize(256),\n",
" transforms.CenterCrop(224),\n",
" transforms.ToTensor(),\n",
" transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
" ]),\n",
"}\n",
"\n",
"data_dir = '/content/gdrive/MyDrive/Colab Notebooks/data/Shuffle Split 10 of 17 Classes Big Brain Tumor MRI Images'\n",
"image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),\n",
" data_transforms[x])\n",
" for x in ['train', 'val']}\n",
"dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=17,\n",
" shuffle=True, num_workers=2)\n",
" for x in ['train', 'val']}\n",
"dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}\n",
"class_names = image_datasets['train'].classes\n",
"\n",
"device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "MFDw_v0wW2lA"
},
"source": [
"### Visualize a few images\n",
"Let's visualize a few training images so as to understand the data\n",
"augmentations.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 147
},
"id": "tceMTgC6W2lB",
"outputId": "a1de05e9-8cf8-46df-de41-e80fd14dc3af"
},
"outputs": [
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
}
],
"source": [
"def imshow(inp, title=None):\n",
" \"\"\"Display image for Tensor.\"\"\"\n",
" inp = inp.numpy().transpose((1, 2, 0))\n",
" mean = np.array([0.485, 0.456, 0.406])\n",
" std = np.array([0.229, 0.224, 0.225])\n",
" inp = std * inp + mean\n",
" inp = np.clip(inp, 0, 1)\n",
" plt.imshow(inp)\n",
" if title is not None:\n",
" plt.title(title)\n",
" plt.pause(0.001) # pause a bit so that plots are updated\n",
"\n",
"\n",
"# Get a batch of training data\n",
"inputs, classes = next(iter(dataloaders['train']))\n",
"\n",
"# Make a grid from batch\n",
"out = torchvision.utils.make_grid(inputs)\n",
"\n",
"imshow(out, title=[class_names[x] for x in classes])"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "mq3LbRiQW2lB"
},
"source": [
"## Training the model\n",
"\n",
"Now, let's write a general function to train a model. Here, we will\n",
"illustrate:\n",
"\n",
"- Scheduling the learning rate\n",
"- Saving the best model\n",
"\n",
"In the following, parameter ``scheduler`` is an LR scheduler object from\n",
"``torch.optim.lr_scheduler``.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"id": "utSazWwYW2lC"
},
"outputs": [],
"source": [
"import time\n",
"import copy\n",
"from sklearn.metrics import confusion_matrix\n",
"\n",
"def train_model(model, criterion, optimizer, scheduler, num_epochs=120):\n",
" since = time.time()\n",
"\n",
" best_model_wts = copy.deepcopy(model.state_dict())\n",
" best_acc = 0.0\n",
"\n",
" for epoch in range(num_epochs):\n",
" print(f'Epoch {epoch}/{num_epochs - 1}')\n",
" print('-' * 10)\n",
"\n",
" # Each epoch has a training and validation phase\n",
" for phase in ['train', 'val']:\n",
" if phase == 'train':\n",
" model.train() # Set model to training mode\n",
" else:\n",
" model.eval() # Set model to evaluate mode\n",
"\n",
" running_loss = 0.0\n",
" running_corrects = 0\n",
" all_labels = []\n",
" all_preds = []\n",
"\n",
" # Iterate over data.\n",
" for inputs, labels in dataloaders[phase]:\n",
" inputs = inputs.to(device)\n",
" labels = labels.to(device)\n",
"\n",
" # zero the parameter gradients\n",
" optimizer.zero_grad()\n",
"\n",
" # forward\n",
" # track history if only in train\n",
" with torch.set_grad_enabled(phase == 'train'):\n",
" outputs = model(inputs)\n",
" _, preds = torch.max(outputs, 1)\n",
" loss = criterion(outputs, labels)\n",
"\n",
" # backward + optimize only if in training phase\n",
" if phase == 'train':\n",
" loss.backward()\n",
" optimizer.step()\n",
"\n",
" # statistics\n",
" running_loss += loss.item() * inputs.size(0)\n",
" running_corrects += torch.sum(preds == labels.data)\n",
" all_labels.extend(labels.data.cpu().numpy())\n",
" all_preds.extend(preds.cpu().numpy())\n",
"\n",
" if phase == 'train':\n",
" scheduler.step()\n",
"\n",
" epoch_loss = running_loss / dataset_sizes[phase]\n",
" epoch_acc = running_corrects.double() / dataset_sizes[phase]\n",
"\n",
" print(f'{phase} Loss: {epoch_loss:.4f} Acc: {epoch_acc:.4f}')\n",
"\n",
" # Calculate confusion matrix\n",
" cm = confusion_matrix(all_labels, all_preds)\n",
" print(f'Confusion Matrix:\\n{cm}')\n",
"\n",
" # deep copy the model\n",
" if phase == 'val' and epoch_acc > best_acc:\n",
" best_acc = epoch_acc\n",
" best_model_wts = copy.deepcopy(model.state_dict())\n",
"\n",
" print()\n",
"\n",
" time_elapsed = time.time() - since\n",
" print(f'Training complete in {time_elapsed // 60:.0f}m {time_elapsed % 60:.0f}s')\n",
" print(f'Best val Acc: {best_acc:4f}')\n",
"\n",
" # load best model weights\n",
" model.load_state_dict(best_model_wts)\n",
" return model"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "h09FnkFVW2lC"
},
"source": [
"### Visualizing the model predictions\n",
"\n",
"Generic function to display predictions for a few images\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {
"id": "MVpCbBziW2lC"
},
"outputs": [],
"source": [
"def visualize_model(model, num_images=6):\n",
" was_training = model.training\n",
" model.eval()\n",
" images_so_far = 0\n",
" fig = plt.figure()\n",
"\n",
" with torch.no_grad():\n",
" for i, (inputs, labels) in enumerate(dataloaders['val']):\n",
" inputs = inputs.to(device)\n",
" labels = labels.to(device)\n",
"\n",
" outputs = model(inputs)\n",
" _, preds = torch.max(outputs, 1)\n",
"\n",
" for j in range(inputs.size()[0]):\n",
" images_so_far += 1\n",
" ax = plt.subplot(num_images//2, 2, images_so_far)\n",
" ax.axis('off')\n",
" ax.set_title(f'predicted: {class_names[preds[j]]}')\n",
" imshow(inputs.cpu().data[j])\n",
"\n",
" if images_so_far == num_images:\n",
" model.train(mode=was_training)\n",
" return\n",
" model.train(mode=was_training)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "LxC7G961W2lD"
},
"source": [
"## Finetuning the ConvNet\n",
"\n",
"Load a pretrained model and reset final fully connected layer.\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {
"id": "d0DTgqMvW2lD"
},
"outputs": [],
"source": [
"model_ft = models.resnet18(weights='IMAGENET1K_V1')\n",
"num_ftrs = model_ft.fc.in_features\n",
"# Here the size of each output sample is set to 2.\n",
"# Alternatively, it can be generalized to ``nn.Linear(num_ftrs, len(class_names))``.\n",
"model_ft.fc = nn.Linear(num_ftrs, 10)\n",
"\n",
"model_ft = model_ft.to(device)\n",
"\n",
"criterion = nn.CrossEntropyLoss()\n",
"\n",
"# Observe that all parameters are being optimized\n",
"optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, momentum=0.93)\n",
"\n",
"# Decay LR by a factor of 0.1 every 7 epochs\n",
"exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)"
]
},
{
"cell_type": "code",
"source": [
"seconds = time.time()\n",
"print(\"Time in seconds since beginning of run:\", seconds)\n",
"local_time = time.ctime(seconds)\n",
"print(local_time)"
],
"metadata": {
"id": "YRwm8Zp116Se",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "02955c42-7391-47ed-b522-f9b89ed9d4ef"
},
"execution_count": 18,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Time in seconds since beginning of run: 1685206886.1344335\n",
"Sat May 27 17:01:26 2023\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "asgAWJjkW2lD"
},
"source": [
"### Train and evaluate\n",
"\n",
"It should take around 15-25 min on CPU. On GPU though, it takes less than a\n",
"minute.\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"id": "RX6GdXXJW2lE",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "1e8ded54-a11c-4154-a828-831170e9db22"
},
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Epoch 0/119\n",
"----------\n",
"train Loss: 1.4830 Acc: 0.4559\n",
"Confusion Matrix:\n",
"[[114 26 2 42 46 0 20 1 7 5]\n",
" [ 16 124 4 10 97 5 6 4 7 8]\n",
" [ 9 9 76 0 15 52 5 37 0 3]\n",
" [ 60 12 4 73 42 5 16 1 3 0]\n",
" [ 25 58 8 13 254 4 3 6 7 14]\n",
" [ 5 11 51 2 22 73 1 37 2 2]\n",
" [ 29 7 5 25 19 2 77 3 2 2]\n",
" [ 1 9 31 1 13 30 0 95 1 1]\n",
" [ 5 24 2 4 45 3 1 0 74 1]\n",
" [ 4 10 1 1 55 0 3 5 1 42]]\n",
"val Loss: 0.7989 Acc: 0.6929\n",
"Confusion Matrix:\n",
"[[ 72 2 3 42 2 0 32 0 0 4]\n",
" [ 3 90 6 5 56 0 0 0 1 6]\n",
" [ 0 0 86 0 0 30 1 5 0 1]\n",
" [ 1 0 2 109 3 0 13 0 0 1]\n",
" [ 1 5 0 16 195 3 3 0 3 7]\n",
" [ 0 0 31 0 1 91 0 0 0 0]\n",
" [ 0 0 0 10 0 0 91 0 0 0]\n",
" [ 0 0 23 0 0 39 0 47 0 0]\n",
" [ 0 4 0 2 6 0 0 0 82 1]\n",
" [ 0 4 4 1 17 1 1 0 0 44]]\n",
"\n",
"Epoch 1/119\n",
"----------\n",
"train Loss: 0.8319 Acc: 0.6865\n",
"Confusion Matrix:\n",
"[[181 10 1 41 6 0 22 0 0 2]\n",
" [ 6 190 1 5 52 3 5 1 4 14]\n",
" [ 3 0 119 2 8 49 2 21 0 2]\n",
" [ 34 6 1 143 13 0 16 0 2 1]\n",
" [ 3 45 7 18 284 1 4 1 13 16]\n",
" [ 1 2 46 1 5 127 0 23 0 1]\n",
" [ 13 3 1 14 3 0 136 1 0 0]\n",
" [ 0 0 28 0 2 19 1 132 0 0]\n",
" [ 1 11 0 5 16 0 0 0 124 2]\n",
" [ 3 12 3 1 28 0 0 1 1 73]]\n",
"val Loss: 0.5628 Acc: 0.8014\n",
"Confusion Matrix:\n",
"[[128 4 0 11 0 0 13 1 0 0]\n",
" [ 1 151 0 2 3 0 0 2 6 2]\n",
" [ 0 0 58 0 3 24 0 37 0 1]\n",
" [ 27 0 0 88 3 1 8 1 1 0]\n",
" [ 0 17 0 7 196 1 1 1 5 5]\n",
" [ 0 0 0 0 1 84 0 38 0 0]\n",
" [ 1 1 0 1 1 0 97 0 0 0]\n",
" [ 0 0 0 0 0 1 0 108 0 0]\n",
" [ 0 2 0 1 3 0 0 0 89 0]\n",
" [ 2 10 0 0 6 0 0 3 1 50]]\n",
"\n",
"Epoch 2/119\n",
"----------\n",
"train Loss: 0.6673 Acc: 0.7652\n",
"Confusion Matrix:\n",
"[[191 5 2 40 7 0 13 2 0 3]\n",
" [ 10 209 0 7 32 0 5 3 5 10]\n",
" [ 2 1 130 0 6 45 1 20 0 1]\n",
" [ 34 0 1 154 11 0 11 2 1 2]\n",
" [ 5 26 2 14 328 3 1 1 5 7]\n",
" [ 0 3 38 0 4 140 0 21 0 0]\n",
" [ 10 2 1 15 0 0 142 0 1 0]\n",
" [ 0 1 15 0 1 8 0 156 0 1]\n",
" [ 1 7 0 2 9 0 0 0 140 0]\n",
" [ 3 5 2 3 15 0 2 0 0 92]]\n",
"val Loss: 0.4355 Acc: 0.8403\n",
"Confusion Matrix:\n",
"[[131 6 0 17 0 0 2 0 1 0]\n",
" [ 0 159 0 0 2 0 0 0 2 4]\n",
" [ 0 1 101 0 0 14 0 6 0 1]\n",
" [ 7 3 1 113 3 0 2 0 0 0]\n",
" [ 0 20 1 4 199 0 0 0 2 7]\n",
" [ 0 0 42 0 2 76 0 1 0 2]\n",
" [ 2 3 0 4 0 0 92 0 0 0]\n",
" [ 0 1 18 0 0 8 0 82 0 0]\n",
" [ 0 5 0 1 0 0 0 0 89 0]\n",
" [ 1 9 0 0 4 0 0 0 0 58]]\n",
"\n",
"Epoch 3/119\n",
"----------\n",
"train Loss: 0.5840 Acc: 0.7830\n",
"Confusion Matrix:\n",
"[[200 4 1 33 6 0 17 0 0 2]\n",
" [ 5 230 1 1 28 1 1 0 6 8]\n",
" [ 1 1 131 3 3 49 0 18 0 0]\n",
" [ 28 0 0 161 11 0 15 0 0 1]\n",
" [ 7 24 3 8 325 3 4 0 10 8]\n",
" [ 0 1 44 1 3 143 0 13 0 1]\n",
" [ 11 1 0 12 6 0 140 0 0 1]\n",
" [ 0 0 14 0 0 16 0 152 0 0]\n",
" [ 2 2 0 4 11 0 0 1 137 2]\n",
" [ 0 9 1 2 6 2 0 0 0 102]]\n",
"val Loss: 0.3604 Acc: 0.8617\n",
"Confusion Matrix:\n",
"[[137 4 0 12 0 0 3 0 1 0]\n",
" [ 0 165 0 0 2 0 0 0 0 0]\n",
" [ 0 1 112 0 0 5 1 4 0 0]\n",
" [ 9 0 0 116 3 0 1 0 0 0]\n",
" [ 0 29 0 2 198 2 1 0 1 0]\n",
" [ 0 2 39 0 1 76 0 5 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 6 0 0 2 1 100 0 0]\n",
" [ 0 15 0 1 1 0 0 0 78 0]\n",
" [ 4 16 1 2 2 0 1 0 0 46]]\n",
"\n",
"Epoch 4/119\n",
"----------\n",
"train Loss: 0.5298 Acc: 0.8062\n",
"Confusion Matrix:\n",
"[[215 4 0 26 2 0 12 0 3 1]\n",
" [ 4 229 0 4 26 0 3 0 6 9]\n",
" [ 1 1 140 1 2 40 0 20 0 1]\n",
" [ 26 4 3 164 8 0 7 2 1 1]\n",
" [ 8 32 2 3 337 4 0 0 0 6]\n",
" [ 1 3 32 1 3 146 1 19 0 0]\n",
" [ 9 0 1 10 2 0 149 0 0 0]\n",
" [ 0 0 7 2 0 12 0 161 0 0]\n",
" [ 1 13 0 2 8 0 0 0 135 0]\n",
" [ 2 8 0 1 12 1 1 0 1 96]]\n",
"val Loss: 0.4613 Acc: 0.8358\n",
"Confusion Matrix:\n",
"[[126 9 0 4 0 0 17 0 1 0]\n",
" [ 0 161 0 0 4 0 0 0 2 0]\n",
" [ 0 4 99 0 0 1 1 17 0 1]\n",
" [ 3 1 0 106 5 0 14 0 0 0]\n",
" [ 0 9 1 1 219 0 0 0 0 3]\n",
" [ 0 3 68 0 7 29 0 14 0 2]\n",
" [ 0 1 0 0 0 0 100 0 0 0]\n",
" [ 0 3 1 0 0 0 0 105 0 0]\n",
" [ 0 2 0 0 3 0 1 0 89 0]\n",
" [ 1 11 0 0 0 0 0 0 0 60]]\n",
"\n",
"Epoch 5/119\n",
"----------\n",
"train Loss: 0.4473 Acc: 0.8408\n",
"Confusion Matrix:\n",
"[[215 6 1 29 2 1 9 0 0 0]\n",
" [ 5 240 1 2 25 0 1 0 4 3]\n",
" [ 0 0 157 3 1 30 0 14 0 1]\n",
" [ 25 1 0 173 8 1 7 0 1 0]\n",
" [ 1 22 3 8 345 2 2 0 3 6]\n",
" [ 1 0 34 0 2 158 0 9 0 2]\n",
" [ 9 1 0 8 3 0 150 0 0 0]\n",
" [ 0 0 5 0 0 11 0 166 0 0]\n",
" [ 3 6 0 2 3 0 2 0 143 0]\n",
" [ 2 4 3 1 10 1 0 0 0 101]]\n",
"val Loss: 0.3165 Acc: 0.8953\n",
"Confusion Matrix:\n",
"[[128 0 1 16 0 0 10 0 2 0]\n",
" [ 0 156 0 1 3 0 0 0 6 1]\n",
" [ 0 0 108 0 0 6 0 9 0 0]\n",
" [ 0 0 0 121 2 0 6 0 0 0]\n",
" [ 0 4 0 4 222 2 0 0 1 0]\n",
" [ 0 0 38 0 1 78 0 6 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 7 0 0 0 0 102 0 0]\n",
" [ 0 1 0 1 2 0 0 0 91 0]\n",
" [ 0 2 0 1 4 0 0 0 0 65]]\n",
"\n",
"Epoch 6/119\n",
"----------\n",
"train Loss: 0.4065 Acc: 0.8681\n",
"Confusion Matrix:\n",
"[[225 6 1 22 1 0 7 0 1 0]\n",
" [ 4 251 0 2 13 1 0 0 4 6]\n",
" [ 0 0 167 1 2 23 0 13 0 0]\n",
" [ 20 2 0 180 9 0 3 0 1 1]\n",
" [ 2 21 2 7 355 1 3 0 0 1]\n",
" [ 0 1 22 0 0 168 0 14 0 1]\n",
" [ 5 1 1 4 1 0 158 0 0 1]\n",
" [ 0 0 7 0 1 11 1 161 0 1]\n",
" [ 0 8 0 0 6 0 1 0 142 2]\n",
" [ 1 3 1 2 12 0 0 1 1 101]]\n",
"val Loss: 0.3090 Acc: 0.8824\n",
"Confusion Matrix:\n",
"[[147 1 0 8 0 0 0 0 1 0]\n",
" [ 0 154 0 0 7 0 0 1 0 5]\n",
" [ 0 0 77 0 0 40 0 6 0 0]\n",
" [ 6 1 0 116 2 0 3 0 0 1]\n",
" [ 1 7 1 1 219 1 0 0 0 3]\n",
" [ 0 0 4 0 1 115 0 3 0 0]\n",
" [ 20 1 0 2 1 0 77 0 0 0]\n",
" [ 0 0 1 0 0 11 0 97 0 0]\n",
" [ 0 3 1 0 2 0 0 0 89 0]\n",
" [ 2 3 0 0 2 1 0 0 0 64]]\n",
"\n",
"Epoch 7/119\n",
"----------\n",
"train Loss: 0.3531 Acc: 0.8708\n",
"Confusion Matrix:\n",
"[[231 3 0 16 1 0 10 0 2 0]\n",
" [ 6 252 0 2 10 0 0 0 5 6]\n",
" [ 0 0 169 1 2 25 0 9 0 0]\n",
" [ 15 1 0 186 3 0 10 1 0 0]\n",
" [ 3 27 4 4 344 2 1 0 5 2]\n",
" [ 0 1 26 0 3 157 0 18 0 1]\n",
" [ 8 1 1 8 1 1 151 0 0 0]\n",
" [ 0 1 10 0 1 4 0 166 0 0]\n",
" [ 0 2 0 2 3 0 1 0 151 0]\n",
" [ 0 4 0 3 7 0 1 0 0 107]]\n",
"val Loss: 0.2069 Acc: 0.9213\n",
"Confusion Matrix:\n",
"[[138 2 0 12 0 0 4 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 101 0 0 14 0 8 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 7 1 2 222 0 0 0 0 1]\n",
" [ 0 0 13 0 1 103 0 6 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 1 0 0 2 0 106 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 2 7 0 1 1 0 1 0 0 60]]\n",
"\n",
"Epoch 8/119\n",
"----------\n",
"train Loss: 0.2991 Acc: 0.8999\n",
"Confusion Matrix:\n",
"[[236 2 0 21 1 0 2 0 1 0]\n",
" [ 7 259 0 1 11 0 1 0 0 2]\n",
" [ 2 0 172 0 0 21 0 10 0 1]\n",
" [ 13 1 1 190 4 0 4 1 1 1]\n",
" [ 2 12 4 0 365 0 3 1 2 3]\n",
" [ 0 0 20 0 1 176 0 9 0 0]\n",
" [ 7 2 0 6 2 0 152 0 2 0]\n",
" [ 0 0 5 0 0 8 0 169 0 0]\n",
" [ 1 3 0 3 4 0 0 0 147 1]\n",
" [ 0 2 0 0 4 2 1 0 1 112]]\n",
"val Loss: 0.1989 Acc: 0.9251\n",
"Confusion Matrix:\n",
"[[143 2 0 6 0 0 5 0 1 0]\n",
" [ 0 161 0 0 2 0 0 0 3 1]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 5 0 0 122 2 0 0 0 0 0]\n",
" [ 0 9 1 3 215 1 0 0 1 3]\n",
" [ 0 0 17 0 1 100 0 5 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 9/119\n",
"----------\n",
"train Loss: 0.3139 Acc: 0.8931\n",
"Confusion Matrix:\n",
"[[233 2 0 16 4 0 6 2 0 0]\n",
" [ 4 258 0 1 13 0 2 0 0 3]\n",
" [ 0 0 169 0 0 28 0 9 0 0]\n",
" [ 10 2 0 192 3 0 7 0 0 2]\n",
" [ 2 14 1 2 362 2 1 1 5 2]\n",
" [ 2 0 19 0 3 166 1 14 0 1]\n",
" [ 7 0 0 5 0 0 157 2 0 0]\n",
" [ 1 1 4 0 1 4 0 171 0 0]\n",
" [ 2 4 0 2 5 0 0 0 146 0]\n",
" [ 1 6 0 1 4 0 0 0 1 109]]\n",
"val Loss: 0.1924 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 102 0 0 14 0 7 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 4 1 2 224 0 0 0 0 2]\n",
" [ 0 0 8 0 2 109 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 1 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 4 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 10/119\n",
"----------\n",
"train Loss: 0.2656 Acc: 0.9095\n",
"Confusion Matrix:\n",
"[[240 0 0 18 1 0 4 0 0 0]\n",
" [ 5 256 0 0 16 0 0 0 2 2]\n",
" [ 2 0 170 0 2 18 2 12 0 0]\n",
" [ 11 0 0 196 3 0 5 1 0 0]\n",
" [ 3 10 0 7 366 2 1 0 0 3]\n",
" [ 0 0 19 0 2 177 0 6 0 2]\n",
" [ 8 0 0 2 0 0 161 0 0 0]\n",
" [ 0 0 7 0 1 6 0 168 0 0]\n",
" [ 0 3 0 1 1 0 0 0 154 0]\n",
" [ 1 2 0 0 6 0 0 0 2 111]]\n",
"val Loss: 0.1888 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 2 1]\n",
" [ 0 0 105 0 0 7 0 11 0 0]\n",
" [ 2 0 0 123 2 0 2 0 0 0]\n",
" [ 0 7 1 2 221 0 0 0 0 2]\n",
" [ 0 0 8 0 3 107 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 1 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 5 0 1 1 0 1 0 0 64]]\n",
"\n",
"Epoch 11/119\n",
"----------\n",
"train Loss: 0.2778 Acc: 0.9040\n",
"Confusion Matrix:\n",
"[[233 3 0 17 1 0 7 0 0 2]\n",
" [ 4 255 0 0 13 1 0 1 2 5]\n",
" [ 0 1 176 0 0 21 1 7 0 0]\n",
" [ 8 3 0 196 1 1 6 0 1 0]\n",
" [ 1 9 0 4 366 0 1 2 3 6]\n",
" [ 0 0 19 2 0 172 1 12 0 0]\n",
" [ 6 1 0 5 1 0 158 0 0 0]\n",
" [ 0 1 8 0 0 2 0 171 0 0]\n",
" [ 1 7 0 1 3 0 0 0 147 0]\n",
" [ 1 0 3 0 4 0 0 1 0 113]]\n",
"val Loss: 0.1888 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[142 2 0 6 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 104 0 0 9 0 10 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 7 1 2 223 0 0 0 0 0]\n",
" [ 0 0 8 0 2 109 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 6 0 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 12/119\n",
"----------\n",
"train Loss: 0.2886 Acc: 0.8967\n",
"Confusion Matrix:\n",
"[[238 4 0 16 2 0 3 0 0 0]\n",
" [ 3 252 0 2 14 1 0 0 2 7]\n",
" [ 0 0 170 0 0 23 0 13 0 0]\n",
" [ 8 2 1 190 8 1 6 0 0 0]\n",
" [ 0 16 1 3 364 1 1 0 2 4]\n",
" [ 1 0 19 0 1 175 1 9 0 0]\n",
" [ 4 0 0 5 0 0 162 0 0 0]\n",
" [ 0 0 7 0 0 5 1 168 0 1]\n",
" [ 0 3 0 0 10 0 0 1 145 0]\n",
" [ 1 5 1 0 6 0 2 0 0 107]]\n",
"val Loss: 0.1903 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[139 2 0 10 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 6 1 3 223 0 0 0 0 0]\n",
" [ 0 0 11 0 2 106 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 13/119\n",
"----------\n",
"train Loss: 0.2715 Acc: 0.9013\n",
"Confusion Matrix:\n",
"[[233 3 0 17 1 0 8 0 0 1]\n",
" [ 5 254 0 1 16 0 1 0 1 3]\n",
" [ 1 1 169 1 0 21 0 12 0 1]\n",
" [ 12 2 0 195 3 0 3 0 0 1]\n",
" [ 2 14 1 2 362 0 1 0 3 7]\n",
" [ 0 0 20 0 0 179 0 7 0 0]\n",
" [ 3 4 0 4 2 0 158 0 0 0]\n",
" [ 1 1 8 0 0 4 0 168 0 0]\n",
" [ 0 7 0 2 3 0 0 0 147 0]\n",
" [ 1 2 0 0 3 0 0 0 0 116]]\n",
"val Loss: 0.1836 Acc: 0.9305\n",
"Confusion Matrix:\n",
"[[146 1 0 5 0 0 5 0 0 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 110 0 0 9 0 4 0 0]\n",
" [ 5 0 0 121 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 19 0 1 99 0 4 0 0]\n",
" [ 1 0 0 0 0 0 100 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 2 2 0 0 0 90 0]\n",
" [ 2 3 2 1 1 0 0 0 0 63]]\n",
"\n",
"Epoch 14/119\n",
"----------\n",
"train Loss: 0.2793 Acc: 0.9049\n",
"Confusion Matrix:\n",
"[[234 1 1 17 3 0 6 0 0 1]\n",
" [ 4 255 0 1 13 0 0 0 3 5]\n",
" [ 0 0 181 0 0 18 0 7 0 0]\n",
" [ 15 1 3 190 3 0 4 0 0 0]\n",
" [ 4 8 2 5 366 2 3 0 0 2]\n",
" [ 0 0 20 0 4 174 0 8 0 0]\n",
" [ 6 1 1 1 2 1 159 0 0 0]\n",
" [ 1 0 6 0 1 7 0 167 0 0]\n",
" [ 0 4 0 2 5 0 0 0 148 0]\n",
" [ 0 0 0 3 4 0 0 0 0 115]]\n",
"val Loss: 0.2062 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[140 1 0 10 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 6 0 12 0 0]\n",
" [ 0 0 0 127 2 0 0 0 0 0]\n",
" [ 0 2 1 4 226 0 0 0 0 0]\n",
" [ 0 0 13 0 2 102 0 6 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 0 0 108 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 2 1 1 0 1 0 0 64]]\n",
"\n",
"Epoch 15/119\n",
"----------\n",
"train Loss: 0.2790 Acc: 0.9108\n",
"Confusion Matrix:\n",
"[[240 3 1 14 0 0 4 0 0 1]\n",
" [ 3 258 0 0 16 0 0 0 1 3]\n",
" [ 1 0 174 0 4 12 0 12 0 3]\n",
" [ 17 2 0 190 5 0 2 0 0 0]\n",
" [ 1 14 3 3 366 2 0 0 2 1]\n",
" [ 0 0 22 0 1 178 0 5 0 0]\n",
" [ 3 1 0 5 0 0 161 1 0 0]\n",
" [ 0 1 5 0 0 8 0 168 0 0]\n",
" [ 0 3 0 1 2 0 0 0 153 0]\n",
" [ 1 1 1 1 4 0 0 0 0 114]]\n",
"val Loss: 0.1757 Acc: 0.9396\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 2 0 0 125 2 0 0 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 9 0 2 109 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 2 0 104 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 0 1 1 0 1 0 0 66]]\n",
"\n",
"Epoch 16/119\n",
"----------\n",
"train Loss: 0.2493 Acc: 0.9131\n",
"Confusion Matrix:\n",
"[[236 3 0 17 1 0 3 1 0 2]\n",
" [ 4 259 0 1 10 0 2 0 2 3]\n",
" [ 1 0 182 1 0 14 0 7 0 1]\n",
" [ 9 2 0 202 1 0 2 0 0 0]\n",
" [ 4 9 1 2 367 1 1 1 4 2]\n",
" [ 0 0 21 1 3 173 0 8 0 0]\n",
" [ 8 1 0 1 1 0 160 0 0 0]\n",
" [ 0 0 6 0 0 5 0 170 0 1]\n",
" [ 1 5 0 2 3 0 0 0 148 0]\n",
" [ 0 5 0 1 4 0 1 0 1 110]]\n",
"val Loss: 0.1894 Acc: 0.9290\n",
"Confusion Matrix:\n",
"[[140 2 0 7 0 0 7 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 6 0 12 0 0]\n",
" [ 4 0 0 123 1 0 1 0 0 0]\n",
" [ 0 5 1 4 222 0 0 0 0 1]\n",
" [ 0 0 15 0 1 102 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 2 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 17/119\n",
"----------\n",
"train Loss: 0.2696 Acc: 0.9040\n",
"Confusion Matrix:\n",
"[[240 2 1 16 1 0 3 0 0 0]\n",
" [ 2 260 0 0 13 0 1 1 1 3]\n",
" [ 0 0 174 1 1 20 0 9 0 1]\n",
" [ 13 2 0 189 5 0 6 0 1 0]\n",
" [ 3 17 2 4 359 0 2 1 2 2]\n",
" [ 0 2 16 1 2 175 0 10 0 0]\n",
" [ 5 2 0 5 1 0 158 0 0 0]\n",
" [ 0 0 5 1 1 10 0 165 0 0]\n",
" [ 0 3 0 0 3 0 0 0 153 0]\n",
" [ 0 5 0 0 3 0 0 0 0 114]]\n",
"val Loss: 0.1894 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 6 0 11 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 4 226 0 0 0 0 0]\n",
" [ 0 0 14 0 1 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 0 0 108 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 2 1 0 0 0 0 0 64]]\n",
"\n",
"Epoch 18/119\n",
"----------\n",
"train Loss: 0.2696 Acc: 0.9017\n",
"Confusion Matrix:\n",
"[[240 6 1 12 1 0 2 0 0 1]\n",
" [ 4 256 0 0 14 0 0 2 1 4]\n",
" [ 2 2 172 1 2 18 0 7 0 2]\n",
" [ 15 0 0 191 7 0 2 1 0 0]\n",
" [ 3 15 1 5 359 2 1 1 4 1]\n",
" [ 0 1 16 0 1 181 0 7 0 0]\n",
" [ 7 0 0 5 0 0 158 0 1 0]\n",
" [ 0 0 5 0 0 8 1 168 0 0]\n",
" [ 0 4 0 1 5 0 1 0 147 1]\n",
" [ 3 4 0 0 3 0 1 0 1 110]]\n",
"val Loss: 0.1790 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[144 1 0 4 0 0 7 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 10 0 6 0 0]\n",
" [ 4 0 0 121 2 0 2 0 0 0]\n",
" [ 0 5 1 2 223 0 0 0 0 2]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 2 0 1 0 0 2 0 0 67]]\n",
"\n",
"Epoch 19/119\n",
"----------\n",
"train Loss: 0.2443 Acc: 0.9208\n",
"Confusion Matrix:\n",
"[[239 2 0 14 1 0 5 1 0 1]\n",
" [ 1 259 0 0 15 0 0 0 4 2]\n",
" [ 0 0 186 0 2 11 1 6 0 0]\n",
" [ 8 0 0 200 5 0 3 0 0 0]\n",
" [ 0 12 0 7 367 3 0 0 2 1]\n",
" [ 0 2 13 0 2 183 0 5 0 1]\n",
" [ 6 0 0 5 1 0 159 0 0 0]\n",
" [ 0 0 2 0 1 5 0 173 0 1]\n",
" [ 0 4 0 3 6 0 0 0 146 0]\n",
" [ 0 2 0 3 4 0 1 0 0 112]]\n",
"val Loss: 0.1819 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 3 1 2 225 0 0 0 1 1]\n",
" [ 0 0 18 0 1 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 1 1 0 0 0 0 0 67]]\n",
"\n",
"Epoch 20/119\n",
"----------\n",
"train Loss: 0.2649 Acc: 0.9145\n",
"Confusion Matrix:\n",
"[[243 3 0 8 4 0 5 0 0 0]\n",
" [ 1 260 0 1 11 0 0 0 2 6]\n",
" [ 0 0 177 0 0 23 0 6 0 0]\n",
" [ 12 2 1 189 7 0 5 0 0 0]\n",
" [ 2 11 0 6 366 0 1 0 0 6]\n",
" [ 0 0 14 2 2 175 0 12 0 1]\n",
" [ 3 1 0 6 0 0 160 1 0 0]\n",
" [ 1 0 4 0 1 5 0 171 0 0]\n",
" [ 0 2 0 0 3 0 0 0 154 0]\n",
" [ 0 2 0 0 3 0 0 0 2 115]]\n",
"val Loss: 0.1787 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[143 2 0 6 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 4 1 2 225 0 0 0 0 1]\n",
" [ 0 0 10 0 2 107 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 21/119\n",
"----------\n",
"train Loss: 0.2346 Acc: 0.9290\n",
"Confusion Matrix:\n",
"[[250 1 0 7 0 1 2 1 0 1]\n",
" [ 2 262 0 2 9 0 0 0 3 3]\n",
" [ 0 0 176 1 0 19 0 10 0 0]\n",
" [ 7 2 0 203 3 1 0 0 0 0]\n",
" [ 3 10 1 5 361 3 2 0 4 3]\n",
" [ 0 1 8 0 0 187 0 10 0 0]\n",
" [ 2 0 0 2 0 0 167 0 0 0]\n",
" [ 1 0 4 0 0 8 0 169 0 0]\n",
" [ 0 5 0 1 1 0 0 0 152 0]\n",
" [ 0 2 0 0 4 0 1 0 0 115]]\n",
"val Loss: 0.1860 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[140 2 0 8 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 1 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 22/119\n",
"----------\n",
"train Loss: 0.2825 Acc: 0.9067\n",
"Confusion Matrix:\n",
"[[235 3 0 15 2 0 5 0 1 2]\n",
" [ 4 254 0 3 9 0 2 0 4 5]\n",
" [ 0 1 172 0 1 24 0 8 0 0]\n",
" [ 15 0 0 192 5 1 3 0 0 0]\n",
" [ 4 8 0 4 369 1 1 0 2 3]\n",
" [ 0 2 11 1 3 179 0 8 0 2]\n",
" [ 5 0 0 4 1 0 161 0 0 0]\n",
" [ 0 0 2 1 0 9 0 170 0 0]\n",
" [ 1 0 0 0 6 0 0 0 152 0]\n",
" [ 0 4 1 2 6 0 0 0 0 109]]\n",
"val Loss: 0.1792 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 106 0 0 6 0 11 0 0]\n",
" [ 4 0 0 123 2 0 0 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 13 0 2 103 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 1 1 1 0 0 0 0 66]]\n",
"\n",
"Epoch 23/119\n",
"----------\n",
"train Loss: 0.2515 Acc: 0.9131\n",
"Confusion Matrix:\n",
"[[245 1 1 9 1 1 3 0 0 2]\n",
" [ 2 259 0 1 13 0 0 0 1 5]\n",
" [ 0 0 170 1 1 20 0 13 0 1]\n",
" [ 10 2 0 199 2 0 3 0 0 0]\n",
" [ 2 11 3 2 369 0 0 0 1 4]\n",
" [ 0 0 16 0 1 182 0 6 0 1]\n",
" [ 3 1 0 5 1 0 161 0 0 0]\n",
" [ 0 1 6 0 0 7 0 168 0 0]\n",
" [ 0 5 0 2 6 0 1 0 145 0]\n",
" [ 0 5 1 1 6 0 0 0 0 109]]\n",
"val Loss: 0.1860 Acc: 0.9412\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 157 0 0 6 0 0 0 3 1]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 0 0 0 127 2 0 0 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 9 0 2 109 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 24/119\n",
"----------\n",
"train Loss: 0.2631 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[242 2 1 12 1 0 4 0 0 1]\n",
" [ 4 258 0 1 12 0 0 0 2 4]\n",
" [ 0 1 179 0 0 19 0 7 0 0]\n",
" [ 12 0 1 191 5 0 6 0 1 0]\n",
" [ 5 13 2 12 350 1 1 0 3 5]\n",
" [ 0 0 17 0 3 178 0 8 0 0]\n",
" [ 4 0 0 3 0 0 163 1 0 0]\n",
" [ 0 0 2 0 0 4 0 176 0 0]\n",
" [ 0 5 1 3 4 0 0 0 146 0]\n",
" [ 0 4 0 0 5 0 0 0 0 113]]\n",
"val Loss: 0.1856 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[141 1 0 10 0 0 4 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 106 0 0 9 0 8 0 0]\n",
" [ 1 0 0 126 1 0 1 0 0 0]\n",
" [ 0 2 1 5 224 0 0 0 0 1]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 1 2 0 1 0 0 1 0 0 67]]\n",
"\n",
"Epoch 25/119\n",
"----------\n",
"train Loss: 0.2599 Acc: 0.9099\n",
"Confusion Matrix:\n",
"[[241 6 0 12 0 0 4 0 0 0]\n",
" [ 3 249 0 0 15 2 1 0 3 8]\n",
" [ 1 2 171 1 2 19 1 9 0 0]\n",
" [ 7 1 0 199 4 0 4 0 0 1]\n",
" [ 2 14 2 3 364 1 1 0 3 2]\n",
" [ 0 0 11 0 1 186 0 8 0 0]\n",
" [ 4 1 0 3 1 0 162 0 0 0]\n",
" [ 0 1 1 0 0 5 0 174 0 1]\n",
" [ 0 6 0 2 3 0 0 0 147 1]\n",
" [ 0 7 0 1 7 0 0 0 0 107]]\n",
"val Loss: 0.1816 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 2 0 0 123 2 0 2 0 0 0]\n",
" [ 0 4 1 4 224 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 26/119\n",
"----------\n",
"train Loss: 0.2422 Acc: 0.9204\n",
"Confusion Matrix:\n",
"[[245 2 1 10 1 0 2 0 1 1]\n",
" [ 5 254 0 1 15 0 0 0 2 4]\n",
" [ 2 1 173 0 3 19 0 8 0 0]\n",
" [ 13 1 0 197 2 0 2 0 1 0]\n",
" [ 2 9 1 6 363 3 1 0 4 3]\n",
" [ 0 0 11 1 3 186 0 5 0 0]\n",
" [ 1 0 0 1 0 0 168 0 0 1]\n",
" [ 0 1 3 0 0 3 0 175 0 0]\n",
" [ 3 1 0 2 3 0 0 0 150 0]\n",
" [ 0 3 0 1 5 0 0 0 1 112]]\n",
"val Loss: 0.1820 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 2 0 5 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 4 1 2 226 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 27/119\n",
"----------\n",
"train Loss: 0.2472 Acc: 0.9186\n",
"Confusion Matrix:\n",
"[[241 5 1 12 1 0 3 0 0 0]\n",
" [ 6 258 0 1 12 1 1 0 2 0]\n",
" [ 0 0 180 0 0 16 0 10 0 0]\n",
" [ 8 1 0 198 3 0 5 0 0 1]\n",
" [ 2 12 1 3 368 2 0 0 2 2]\n",
" [ 0 0 16 0 0 179 0 10 0 1]\n",
" [ 7 0 0 5 1 0 158 0 0 0]\n",
" [ 0 0 5 0 0 7 0 170 0 0]\n",
" [ 0 4 0 0 2 0 0 0 153 0]\n",
" [ 1 3 0 1 3 0 0 0 0 114]]\n",
"val Loss: 0.1820 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[144 1 0 5 0 0 6 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 7 1 2 222 0 0 0 1 0]\n",
" [ 0 0 11 0 2 106 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 28/119\n",
"----------\n",
"train Loss: 0.2349 Acc: 0.9268\n",
"Confusion Matrix:\n",
"[[247 2 0 8 0 0 6 0 0 0]\n",
" [ 1 254 1 0 16 0 1 0 4 4]\n",
" [ 1 0 177 0 0 16 0 11 0 1]\n",
" [ 8 1 0 201 3 0 2 0 0 1]\n",
" [ 2 9 1 5 368 1 3 0 0 3]\n",
" [ 0 0 9 0 2 186 0 9 0 0]\n",
" [ 3 2 0 4 0 0 162 0 0 0]\n",
" [ 0 0 6 0 0 2 0 174 0 0]\n",
" [ 0 2 0 1 0 0 0 0 155 1]\n",
" [ 0 2 0 1 4 0 1 1 0 113]]\n",
"val Loss: 0.1781 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 4 1 3 225 0 0 0 0 0]\n",
" [ 0 0 12 0 2 106 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 1 4 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 29/119\n",
"----------\n",
"train Loss: 0.2610 Acc: 0.9099\n",
"Confusion Matrix:\n",
"[[236 4 0 11 3 0 8 1 0 0]\n",
" [ 3 258 0 2 14 0 0 0 0 4]\n",
" [ 0 0 179 0 2 17 0 8 0 0]\n",
" [ 9 1 0 199 3 1 3 0 0 0]\n",
" [ 3 8 3 1 370 0 3 0 2 2]\n",
" [ 0 1 15 1 4 178 0 7 0 0]\n",
" [ 5 1 0 7 1 0 155 0 1 1]\n",
" [ 0 0 6 0 0 9 0 167 0 0]\n",
" [ 0 5 0 3 1 0 0 0 150 0]\n",
" [ 1 5 0 1 6 0 1 0 0 108]]\n",
"val Loss: 0.1773 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[144 1 0 7 0 0 4 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 11 0 5 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 6 1 2 224 0 0 0 0 0]\n",
" [ 0 0 13 0 2 105 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 30/119\n",
"----------\n",
"train Loss: 0.2652 Acc: 0.9090\n",
"Confusion Matrix:\n",
"[[244 2 0 9 0 0 6 0 1 1]\n",
" [ 3 260 0 1 8 0 0 2 1 6]\n",
" [ 1 2 170 0 1 22 0 8 0 2]\n",
" [ 14 0 1 191 4 0 5 0 0 1]\n",
" [ 1 13 2 2 364 2 0 0 2 6]\n",
" [ 0 1 12 0 0 180 0 13 0 0]\n",
" [ 4 0 0 8 2 0 157 0 0 0]\n",
" [ 0 1 2 0 0 10 0 169 0 0]\n",
" [ 0 3 0 1 4 0 0 0 151 0]\n",
" [ 0 7 0 1 2 0 0 0 0 112]]\n",
"val Loss: 0.1845 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[140 1 0 8 0 0 7 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 5 1 3 224 0 0 0 0 0]\n",
" [ 0 0 15 0 1 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 31/119\n",
"----------\n",
"train Loss: 0.2565 Acc: 0.9054\n",
"Confusion Matrix:\n",
"[[240 1 1 11 1 0 9 0 0 0]\n",
" [ 5 258 2 0 8 1 1 0 2 4]\n",
" [ 1 1 173 1 1 22 1 6 0 0]\n",
" [ 12 1 0 194 6 0 2 0 1 0]\n",
" [ 2 14 1 6 363 2 0 0 2 2]\n",
" [ 0 0 12 2 3 177 0 11 0 1]\n",
" [ 4 0 0 2 1 0 163 0 0 1]\n",
" [ 0 1 6 0 0 6 0 169 0 0]\n",
" [ 3 2 0 2 3 0 0 0 149 0]\n",
" [ 1 8 1 1 4 0 0 2 1 104]]\n",
"val Loss: 0.1816 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 163 0 0 1 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 4 0 0 121 2 0 2 0 0 0]\n",
" [ 0 8 1 2 221 0 0 0 0 1]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 32/119\n",
"----------\n",
"train Loss: 0.2469 Acc: 0.9140\n",
"Confusion Matrix:\n",
"[[236 1 0 14 1 0 9 0 0 2]\n",
" [ 4 259 0 2 11 0 0 2 2 1]\n",
" [ 0 1 174 0 0 23 0 7 0 1]\n",
" [ 12 2 0 195 4 0 1 0 0 2]\n",
" [ 0 18 0 1 366 2 0 0 4 1]\n",
" [ 0 0 16 0 0 178 0 12 0 0]\n",
" [ 3 0 0 2 0 0 166 0 0 0]\n",
" [ 0 0 2 0 1 7 0 172 0 0]\n",
" [ 0 4 0 2 3 0 1 0 149 0]\n",
" [ 0 5 1 1 1 0 0 0 0 114]]\n",
"val Loss: 0.1813 Acc: 0.9412\n",
"Confusion Matrix:\n",
"[[140 2 0 9 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 0 0 0 127 2 0 0 0 0 0]\n",
" [ 0 3 0 2 227 1 0 0 0 0]\n",
" [ 0 0 8 0 2 110 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 2 0 105 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 6 0 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 33/119\n",
"----------\n",
"train Loss: 0.2448 Acc: 0.9136\n",
"Confusion Matrix:\n",
"[[237 1 0 18 3 0 2 0 0 2]\n",
" [ 1 258 0 1 14 0 2 1 1 3]\n",
" [ 0 1 172 1 1 22 1 7 0 1]\n",
" [ 9 0 0 198 4 0 2 1 0 2]\n",
" [ 1 17 1 4 361 2 1 0 1 4]\n",
" [ 1 0 11 0 3 185 0 6 0 0]\n",
" [ 1 0 0 5 1 0 164 0 0 0]\n",
" [ 0 0 6 0 0 5 0 171 0 0]\n",
" [ 0 6 0 2 1 0 0 0 150 0]\n",
" [ 2 2 0 0 5 0 1 0 0 112]]\n",
"val Loss: 0.1827 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 10 0 6 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 4 1 4 224 0 0 0 0 0]\n",
" [ 0 0 10 0 2 107 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 34/119\n",
"----------\n",
"train Loss: 0.2489 Acc: 0.9190\n",
"Confusion Matrix:\n",
"[[239 1 0 11 2 0 8 0 0 2]\n",
" [ 4 257 0 0 16 1 0 0 1 2]\n",
" [ 0 2 176 0 0 12 0 14 1 1]\n",
" [ 9 2 0 198 5 0 2 0 0 0]\n",
" [ 0 9 2 2 372 2 0 0 3 2]\n",
" [ 0 0 9 0 1 183 1 11 1 0]\n",
" [ 2 0 0 3 0 0 166 0 0 0]\n",
" [ 0 0 9 0 0 4 0 169 0 0]\n",
" [ 1 3 0 1 4 0 0 0 150 0]\n",
" [ 1 3 1 1 5 0 0 0 1 110]]\n",
"val Loss: 0.1869 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 10 0 7 0 0]\n",
" [ 2 0 0 125 1 0 1 0 0 0]\n",
" [ 0 5 1 4 222 0 0 0 1 0]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 1 0 105 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 35/119\n",
"----------\n",
"train Loss: 0.2567 Acc: 0.9104\n",
"Confusion Matrix:\n",
"[[239 3 1 10 5 0 4 0 0 1]\n",
" [ 4 253 0 0 15 1 0 0 2 6]\n",
" [ 0 0 166 1 1 25 0 13 0 0]\n",
" [ 10 1 1 197 5 0 2 0 0 0]\n",
" [ 0 11 3 2 370 1 0 0 2 3]\n",
" [ 0 0 9 0 1 184 0 12 0 0]\n",
" [ 4 0 0 4 1 0 162 0 0 0]\n",
" [ 0 0 6 1 0 4 1 170 0 0]\n",
" [ 0 3 0 1 6 0 0 0 149 0]\n",
" [ 1 2 0 0 8 0 0 0 0 111]]\n",
"val Loss: 0.1774 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[143 2 0 5 0 0 6 0 1 0]\n",
" [ 0 158 0 0 5 0 0 0 3 1]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 0 1 2 0 0 0 0 66]]\n",
"\n",
"Epoch 36/119\n",
"----------\n",
"train Loss: 0.2642 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[234 5 0 13 2 0 8 0 0 1]\n",
" [ 2 258 1 1 14 0 0 0 2 3]\n",
" [ 0 0 177 0 0 20 0 8 0 1]\n",
" [ 10 3 1 189 3 1 8 0 1 0]\n",
" [ 1 10 2 1 373 0 2 0 1 2]\n",
" [ 1 0 19 0 0 171 1 14 0 0]\n",
" [ 5 0 0 2 2 0 162 0 0 0]\n",
" [ 0 1 3 0 1 6 2 169 0 0]\n",
" [ 0 1 0 1 2 0 0 0 155 0]\n",
" [ 2 3 0 1 7 0 1 0 0 108]]\n",
"val Loss: 0.1900 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[140 2 0 9 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 104 0 0 7 0 12 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 4 1 3 224 0 0 0 1 0]\n",
" [ 0 0 10 0 2 105 0 6 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 0 0 108 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 37/119\n",
"----------\n",
"train Loss: 0.2464 Acc: 0.9167\n",
"Confusion Matrix:\n",
"[[240 2 0 12 0 0 7 0 0 2]\n",
" [ 4 255 0 1 13 0 1 1 1 5]\n",
" [ 1 0 178 0 1 18 1 7 0 0]\n",
" [ 10 0 1 191 3 0 10 0 1 0]\n",
" [ 1 10 0 0 369 2 0 0 4 6]\n",
" [ 0 0 12 0 1 183 0 10 0 0]\n",
" [ 5 0 0 2 1 0 163 0 0 0]\n",
" [ 0 0 3 0 0 5 0 174 0 0]\n",
" [ 0 3 1 1 1 0 0 0 153 0]\n",
" [ 0 4 0 2 6 0 1 0 0 109]]\n",
"val Loss: 0.1798 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 2 0 6 0 0 6 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 5 1 1 224 0 0 0 0 2]\n",
" [ 0 0 14 0 2 102 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 38/119\n",
"----------\n",
"train Loss: 0.2456 Acc: 0.9167\n",
"Confusion Matrix:\n",
"[[243 0 0 13 0 0 3 0 1 3]\n",
" [ 4 268 0 0 4 1 0 0 3 1]\n",
" [ 0 0 171 1 2 23 0 8 0 1]\n",
" [ 12 2 0 191 2 0 7 0 1 1]\n",
" [ 2 12 3 5 364 1 1 0 3 1]\n",
" [ 0 0 14 1 1 177 0 13 0 0]\n",
" [ 6 1 0 2 0 0 162 0 0 0]\n",
" [ 0 0 3 0 0 3 0 176 0 0]\n",
" [ 0 2 0 0 3 0 1 0 153 0]\n",
" [ 1 2 0 2 5 1 0 0 1 110]]\n",
"val Loss: 0.1815 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[143 1 0 6 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 124 1 0 2 0 0 0]\n",
" [ 0 3 1 4 224 0 0 0 0 1]\n",
" [ 0 0 18 0 1 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 39/119\n",
"----------\n",
"train Loss: 0.2573 Acc: 0.9163\n",
"Confusion Matrix:\n",
"[[237 4 0 13 2 0 6 0 1 0]\n",
" [ 1 261 0 1 11 1 1 0 2 3]\n",
" [ 0 0 174 0 1 22 0 9 0 0]\n",
" [ 11 3 0 196 1 0 5 0 0 0]\n",
" [ 1 13 0 3 369 2 1 0 1 2]\n",
" [ 0 0 9 0 3 183 0 10 1 0]\n",
" [ 6 0 0 8 1 0 156 0 0 0]\n",
" [ 0 0 4 0 0 6 0 171 0 1]\n",
" [ 1 1 0 0 5 0 0 0 152 0]\n",
" [ 1 0 0 2 4 0 0 0 0 115]]\n",
"val Loss: 0.1827 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[142 1 0 6 0 0 7 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 108 0 0 10 0 5 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 3 1 6 221 0 0 0 0 2]\n",
" [ 0 0 17 0 1 102 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 40/119\n",
"----------\n",
"train Loss: 0.2549 Acc: 0.9154\n",
"Confusion Matrix:\n",
"[[240 4 1 8 2 0 7 0 0 1]\n",
" [ 3 259 1 0 9 0 2 0 1 6]\n",
" [ 0 1 176 0 0 17 0 12 0 0]\n",
" [ 6 2 0 200 4 0 2 0 2 0]\n",
" [ 1 11 2 1 375 0 0 0 0 2]\n",
" [ 0 0 20 0 0 176 0 10 0 0]\n",
" [ 5 1 0 7 0 0 157 1 0 0]\n",
" [ 0 2 6 0 0 9 0 165 0 0]\n",
" [ 0 2 0 1 3 0 1 0 152 0]\n",
" [ 1 3 1 0 4 0 1 0 0 112]]\n",
"val Loss: 0.1751 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[143 2 0 6 0 0 5 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 3 0]\n",
" [ 0 0 109 0 0 10 0 4 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 8 1 2 221 0 0 0 0 1]\n",
" [ 0 0 15 0 2 103 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 41/119\n",
"----------\n",
"train Loss: 0.2210 Acc: 0.9254\n",
"Confusion Matrix:\n",
"[[238 2 0 16 0 0 6 0 0 1]\n",
" [ 2 265 1 0 9 0 0 0 2 2]\n",
" [ 1 0 180 0 0 18 0 7 0 0]\n",
" [ 6 0 0 202 2 0 4 0 1 1]\n",
" [ 1 10 0 1 377 1 0 0 0 2]\n",
" [ 1 0 12 0 0 182 0 11 0 0]\n",
" [ 3 1 0 5 4 0 158 0 0 0]\n",
" [ 0 1 6 0 0 4 1 169 0 1]\n",
" [ 0 5 0 1 4 0 0 0 149 0]\n",
" [ 0 3 1 1 3 0 0 0 0 114]]\n",
"val Loss: 0.1854 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 6 0 10 0 0]\n",
" [ 3 0 0 122 2 0 2 0 0 0]\n",
" [ 0 3 1 3 226 0 0 0 0 0]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 5 0 1 0 0 2 0 0 63]]\n",
"\n",
"Epoch 42/119\n",
"----------\n",
"train Loss: 0.2662 Acc: 0.9040\n",
"Confusion Matrix:\n",
"[[237 3 0 18 0 0 5 0 0 0]\n",
" [ 4 257 1 2 12 0 0 1 1 3]\n",
" [ 0 0 177 0 0 19 0 9 0 1]\n",
" [ 7 1 0 200 4 0 3 0 1 0]\n",
" [ 1 12 0 5 362 2 1 0 3 6]\n",
" [ 0 0 21 0 2 171 0 11 1 0]\n",
" [ 7 0 0 5 0 0 157 2 0 0]\n",
" [ 0 0 2 1 0 7 0 172 0 0]\n",
" [ 0 5 0 5 2 0 0 0 147 0]\n",
" [ 0 5 0 1 5 0 1 1 2 107]]\n",
"val Loss: 0.1816 Acc: 0.9396\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 105 0 0 9 0 9 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 2 0 1 1 0 2 0 0 66]]\n",
"\n",
"Epoch 43/119\n",
"----------\n",
"train Loss: 0.2711 Acc: 0.9099\n",
"Confusion Matrix:\n",
"[[232 2 0 14 5 0 9 1 0 0]\n",
" [ 4 257 0 0 13 0 0 0 3 4]\n",
" [ 2 1 170 0 1 21 0 11 0 0]\n",
" [ 12 0 0 197 4 1 2 0 0 0]\n",
" [ 1 8 0 3 375 1 1 0 0 3]\n",
" [ 0 2 10 0 2 180 0 12 0 0]\n",
" [ 5 2 0 4 0 0 160 0 0 0]\n",
" [ 0 0 2 0 1 7 0 172 0 0]\n",
" [ 0 5 0 2 3 0 0 0 149 0]\n",
" [ 0 1 1 1 8 3 0 0 0 108]]\n",
"val Loss: 0.1841 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[141 2 0 6 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 13 0 3 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 44/119\n",
"----------\n",
"train Loss: 0.2592 Acc: 0.9113\n",
"Confusion Matrix:\n",
"[[237 4 0 12 3 0 5 0 1 1]\n",
" [ 4 261 0 1 11 0 1 0 2 1]\n",
" [ 1 0 174 1 0 21 1 8 0 0]\n",
" [ 10 0 0 194 7 1 3 0 0 1]\n",
" [ 4 14 1 1 367 4 0 0 1 0]\n",
" [ 0 0 14 0 2 176 0 14 0 0]\n",
" [ 4 0 0 3 1 0 163 0 0 0]\n",
" [ 0 0 5 0 0 2 1 173 0 1]\n",
" [ 0 2 0 0 6 0 1 0 149 1]\n",
" [ 1 5 0 2 4 0 0 0 1 109]]\n",
"val Loss: 0.1861 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[143 1 0 6 0 0 6 0 1 0]\n",
" [ 0 161 0 0 2 0 0 0 3 1]\n",
" [ 0 0 105 0 0 7 0 11 0 0]\n",
" [ 3 0 0 124 1 0 1 0 0 0]\n",
" [ 0 6 1 5 218 0 0 0 1 2]\n",
" [ 0 0 16 0 1 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 45/119\n",
"----------\n",
"train Loss: 0.2346 Acc: 0.9213\n",
"Confusion Matrix:\n",
"[[244 2 0 9 2 0 5 1 0 0]\n",
" [ 4 259 1 2 6 0 1 0 1 7]\n",
" [ 0 1 172 0 1 26 1 5 0 0]\n",
" [ 10 1 0 199 3 0 3 0 0 0]\n",
" [ 0 7 0 3 372 1 2 0 1 6]\n",
" [ 0 0 10 0 1 184 1 10 0 0]\n",
" [ 5 1 0 3 1 0 160 0 0 1]\n",
" [ 0 0 4 0 0 5 0 173 0 0]\n",
" [ 0 4 0 2 2 0 0 0 150 1]\n",
" [ 0 3 0 0 7 0 0 0 0 112]]\n",
"val Loss: 0.1865 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 8 0 8 0 0]\n",
" [ 1 0 0 124 2 0 2 0 0 0]\n",
" [ 0 3 1 3 224 0 0 0 0 2]\n",
" [ 0 0 17 0 1 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 3 0 1 1 0 2 0 0 65]]\n",
"\n",
"Epoch 46/119\n",
"----------\n",
"train Loss: 0.2408 Acc: 0.9199\n",
"Confusion Matrix:\n",
"[[244 3 0 11 1 0 3 0 1 0]\n",
" [ 2 265 1 0 8 0 1 1 0 3]\n",
" [ 0 0 183 0 0 13 0 9 0 1]\n",
" [ 12 3 0 193 3 0 3 0 1 1]\n",
" [ 1 13 2 8 360 3 1 0 1 3]\n",
" [ 0 0 12 0 0 191 0 2 0 1]\n",
" [ 3 1 0 3 2 0 161 0 0 1]\n",
" [ 0 0 7 0 0 5 0 170 0 0]\n",
" [ 0 2 0 3 8 0 0 1 145 0]\n",
" [ 0 3 1 0 7 1 0 0 0 110]]\n",
"val Loss: 0.1822 Acc: 0.9442\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 11 0 5 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 0 1 230 0 0 0 0 0]\n",
" [ 0 0 8 0 2 110 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 4 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 2 0 0 0 0 66]]\n",
"\n",
"Epoch 47/119\n",
"----------\n",
"train Loss: 0.2755 Acc: 0.9067\n",
"Confusion Matrix:\n",
"[[237 1 1 14 1 0 8 0 0 1]\n",
" [ 3 263 1 2 10 0 1 0 0 1]\n",
" [ 0 0 177 1 2 19 0 5 0 2]\n",
" [ 8 1 0 196 8 0 3 0 0 0]\n",
" [ 3 15 2 6 358 1 0 0 1 6]\n",
" [ 0 1 18 0 4 173 0 10 0 0]\n",
" [ 5 0 0 2 1 0 162 1 0 0]\n",
" [ 0 1 4 0 0 9 0 168 0 0]\n",
" [ 0 1 0 2 5 0 1 0 149 1]\n",
" [ 0 1 0 2 5 0 1 0 3 110]]\n",
"val Loss: 0.1827 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[142 2 0 5 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 7 0 9 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 5 1 1 226 0 0 0 0 0]\n",
" [ 0 0 17 0 2 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 2 1 0 0 0 0 0 64]]\n",
"\n",
"Epoch 48/119\n",
"----------\n",
"train Loss: 0.2567 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[244 3 0 10 1 0 4 0 1 0]\n",
" [ 2 257 0 4 12 0 0 0 1 5]\n",
" [ 1 1 179 0 1 17 0 7 0 0]\n",
" [ 12 3 1 190 4 1 4 1 0 0]\n",
" [ 2 13 2 4 363 2 1 0 1 4]\n",
" [ 0 0 18 0 0 173 0 15 0 0]\n",
" [ 4 0 0 5 0 0 162 0 0 0]\n",
" [ 0 0 6 0 0 8 0 168 0 0]\n",
" [ 0 4 0 3 3 0 0 0 149 0]\n",
" [ 0 5 0 0 3 0 1 1 1 111]]\n",
"val Loss: 0.1816 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 3 0]\n",
" [ 0 0 106 0 0 9 0 8 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 9 1 2 219 0 0 0 1 1]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 49/119\n",
"----------\n",
"train Loss: 0.2491 Acc: 0.9177\n",
"Confusion Matrix:\n",
"[[249 3 0 6 0 0 4 0 0 1]\n",
" [ 1 261 0 0 10 0 0 0 4 5]\n",
" [ 0 0 180 0 0 18 0 8 0 0]\n",
" [ 10 0 0 193 6 1 6 0 0 0]\n",
" [ 3 11 3 6 363 0 2 0 2 2]\n",
" [ 0 0 21 0 0 178 0 7 0 0]\n",
" [ 3 0 0 3 1 0 164 0 0 0]\n",
" [ 0 0 7 0 0 2 0 173 0 0]\n",
" [ 1 3 0 1 5 0 0 0 149 0]\n",
" [ 0 5 0 1 8 0 0 1 0 107]]\n",
"val Loss: 0.1795 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[144 1 0 7 0 0 4 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 107 0 0 10 0 6 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 8 1 4 220 0 0 0 0 0]\n",
" [ 0 0 14 0 1 105 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 2 4 1 1 0 0 0 0 0 64]]\n",
"\n",
"Epoch 50/119\n",
"----------\n",
"train Loss: 0.2579 Acc: 0.9172\n",
"Confusion Matrix:\n",
"[[242 2 0 13 1 0 4 0 0 1]\n",
" [ 4 258 0 1 11 0 1 0 3 3]\n",
" [ 0 0 182 1 0 14 1 8 0 0]\n",
" [ 12 1 0 193 4 2 4 0 0 0]\n",
" [ 3 10 1 2 364 1 4 1 2 4]\n",
" [ 0 0 10 0 0 183 1 12 0 0]\n",
" [ 4 0 0 4 0 0 163 0 0 0]\n",
" [ 0 0 6 0 0 6 0 169 0 1]\n",
" [ 0 6 0 1 3 0 1 0 148 0]\n",
" [ 1 5 0 0 2 0 0 0 0 114]]\n",
"val Loss: 0.1846 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 9 0 8 0 0]\n",
" [ 1 0 0 126 1 0 1 0 0 0]\n",
" [ 0 3 1 6 222 0 0 0 0 1]\n",
" [ 0 0 12 0 1 106 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 2 3 0 1 1 0 1 0 0 64]]\n",
"\n",
"Epoch 51/119\n",
"----------\n",
"train Loss: 0.2338 Acc: 0.9240\n",
"Confusion Matrix:\n",
"[[240 0 0 16 1 0 5 0 0 1]\n",
" [ 3 261 0 1 10 1 0 0 2 3]\n",
" [ 0 0 172 0 1 20 1 11 0 1]\n",
" [ 9 0 1 200 4 0 2 0 0 0]\n",
" [ 4 9 3 3 369 1 0 0 1 2]\n",
" [ 1 0 9 0 1 186 0 9 0 0]\n",
" [ 4 0 1 3 0 0 163 0 0 0]\n",
" [ 0 0 3 0 0 5 0 174 0 0]\n",
" [ 0 3 0 2 0 0 0 0 152 2]\n",
" [ 1 2 0 1 4 0 0 0 0 114]]\n",
"val Loss: 0.1818 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 1 0 8 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 2 0 0 125 2 0 0 0 0 0]\n",
" [ 0 3 1 4 223 0 0 0 0 2]\n",
" [ 0 0 14 0 1 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 2 1 1 0 0 0 0 0 68]]\n",
"\n",
"Epoch 52/119\n",
"----------\n",
"train Loss: 0.2509 Acc: 0.9117\n",
"Confusion Matrix:\n",
"[[237 1 0 15 3 0 4 0 0 3]\n",
" [ 3 260 2 1 13 1 0 0 0 1]\n",
" [ 1 0 171 1 1 23 0 8 0 1]\n",
" [ 12 0 0 195 3 1 5 0 0 0]\n",
" [ 0 11 1 2 364 3 2 1 4 4]\n",
" [ 0 0 13 0 2 176 0 15 0 0]\n",
" [ 4 2 0 3 0 0 162 0 0 0]\n",
" [ 0 0 4 0 0 3 0 175 0 0]\n",
" [ 1 3 0 1 3 0 0 0 151 0]\n",
" [ 0 6 1 0 2 0 0 0 0 113]]\n",
"val Loss: 0.1831 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 6 0 12 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 4 1 3 224 0 0 0 1 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 53/119\n",
"----------\n",
"train Loss: 0.2491 Acc: 0.9227\n",
"Confusion Matrix:\n",
"[[244 1 0 11 0 0 7 0 0 0]\n",
" [ 2 265 0 1 8 0 1 0 2 2]\n",
" [ 1 0 180 0 0 18 0 5 0 2]\n",
" [ 9 2 2 192 5 1 2 1 0 2]\n",
" [ 1 10 1 3 368 1 1 0 3 4]\n",
" [ 0 0 12 0 0 184 0 10 0 0]\n",
" [ 7 1 0 2 1 0 160 0 0 0]\n",
" [ 1 0 3 0 0 5 0 173 0 0]\n",
" [ 0 2 0 1 3 0 1 0 152 0]\n",
" [ 1 6 0 0 5 0 0 0 0 110]]\n",
"val Loss: 0.1820 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 7 1 2 223 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 54/119\n",
"----------\n",
"train Loss: 0.2334 Acc: 0.9263\n",
"Confusion Matrix:\n",
"[[244 4 0 8 1 0 3 1 0 2]\n",
" [ 6 263 0 0 6 0 1 0 2 3]\n",
" [ 0 0 184 0 0 15 0 7 0 0]\n",
" [ 9 0 0 200 6 0 1 0 0 0]\n",
" [ 2 7 3 7 368 1 0 0 3 1]\n",
" [ 0 0 17 0 2 176 0 10 0 1]\n",
" [ 2 0 0 3 0 0 166 0 0 0]\n",
" [ 0 0 6 0 1 3 1 171 0 0]\n",
" [ 0 3 0 2 3 0 0 0 151 0]\n",
" [ 0 4 0 0 5 0 0 0 0 113]]\n",
"val Loss: 0.1838 Acc: 0.9305\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 104 0 0 8 0 11 0 0]\n",
" [ 3 0 0 123 1 0 2 0 0 0]\n",
" [ 0 3 1 5 223 0 0 0 0 1]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 0 3 0 1 1 0 2 0 0 65]]\n",
"\n",
"Epoch 55/119\n",
"----------\n",
"train Loss: 0.2478 Acc: 0.9167\n",
"Confusion Matrix:\n",
"[[241 4 0 13 1 0 4 0 0 0]\n",
" [ 2 258 0 0 14 0 1 0 3 3]\n",
" [ 0 0 179 1 2 16 0 7 0 1]\n",
" [ 7 2 0 198 5 0 3 0 1 0]\n",
" [ 3 13 1 4 364 3 1 0 2 1]\n",
" [ 0 0 17 0 3 176 0 10 0 0]\n",
" [ 8 0 0 1 0 1 160 0 1 0]\n",
" [ 0 1 2 0 0 1 0 178 0 0]\n",
" [ 0 3 0 2 4 0 0 0 149 1]\n",
" [ 0 4 0 0 4 0 1 0 1 112]]\n",
"val Loss: 0.1881 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 3 0]\n",
" [ 0 0 105 0 0 7 0 11 0 0]\n",
" [ 3 0 0 122 2 0 2 0 0 0]\n",
" [ 0 7 1 4 219 0 0 0 1 1]\n",
" [ 0 0 13 0 2 103 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 2 0 0 64]]\n",
"\n",
"Epoch 56/119\n",
"----------\n",
"train Loss: 0.2654 Acc: 0.9186\n",
"Confusion Matrix:\n",
"[[243 5 0 10 0 0 4 1 0 0]\n",
" [ 2 257 0 1 13 2 0 0 1 5]\n",
" [ 1 2 181 1 0 16 0 5 0 0]\n",
" [ 9 2 0 199 4 0 2 0 0 0]\n",
" [ 1 18 1 1 363 3 1 1 2 1]\n",
" [ 1 0 15 1 1 183 0 5 0 0]\n",
" [ 4 1 0 4 0 0 161 0 0 1]\n",
" [ 0 1 3 0 1 5 0 172 0 0]\n",
" [ 0 1 0 1 5 0 0 0 152 0]\n",
" [ 0 3 1 2 7 1 0 0 0 108]]\n",
"val Loss: 0.1884 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[138 2 0 9 0 0 7 0 1 0]\n",
" [ 0 156 0 0 6 0 0 0 3 2]\n",
" [ 0 0 106 0 0 10 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 2 226 0 0 0 0 2]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 0 0 0 0 0 68]]\n",
"\n",
"Epoch 57/119\n",
"----------\n",
"train Loss: 0.2538 Acc: 0.9204\n",
"Confusion Matrix:\n",
"[[242 2 0 9 3 0 7 0 0 0]\n",
" [ 4 254 1 2 14 0 0 0 1 5]\n",
" [ 2 0 178 0 3 18 0 5 0 0]\n",
" [ 10 1 1 194 5 0 3 2 0 0]\n",
" [ 1 9 0 1 379 1 1 0 0 0]\n",
" [ 0 0 12 0 0 180 0 14 0 0]\n",
" [ 1 1 0 5 0 0 164 0 0 0]\n",
" [ 0 0 3 0 0 6 0 172 0 1]\n",
" [ 1 4 0 0 3 0 1 0 150 0]\n",
" [ 1 3 1 1 5 0 1 0 0 110]]\n",
"val Loss: 0.1798 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[141 1 0 8 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 10 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 3 1 4 224 0 0 0 0 1]\n",
" [ 0 0 13 0 2 105 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 58/119\n",
"----------\n",
"train Loss: 0.2497 Acc: 0.9190\n",
"Confusion Matrix:\n",
"[[239 1 0 16 1 0 6 0 0 0]\n",
" [ 2 262 0 1 9 0 0 1 1 5]\n",
" [ 1 0 179 0 0 13 1 12 0 0]\n",
" [ 9 2 1 198 4 0 2 0 0 0]\n",
" [ 4 11 1 5 364 3 1 1 2 0]\n",
" [ 0 0 12 1 1 185 0 7 0 0]\n",
" [ 3 1 0 4 0 0 163 0 0 0]\n",
" [ 0 0 9 0 0 4 0 168 0 1]\n",
" [ 0 5 0 1 3 0 0 0 150 0]\n",
" [ 0 4 0 0 4 0 1 0 1 112]]\n",
"val Loss: 0.1810 Acc: 0.9396\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 59/119\n",
"----------\n",
"train Loss: 0.2505 Acc: 0.9181\n",
"Confusion Matrix:\n",
"[[241 5 0 13 2 0 2 0 0 0]\n",
" [ 7 255 0 0 11 1 0 1 0 6]\n",
" [ 1 0 176 1 1 13 0 14 0 0]\n",
" [ 10 0 0 197 2 1 5 1 0 0]\n",
" [ 0 10 1 5 370 2 0 0 3 1]\n",
" [ 0 1 6 0 1 191 0 6 0 1]\n",
" [ 3 1 0 3 0 0 164 0 0 0]\n",
" [ 0 0 4 0 0 7 0 171 0 0]\n",
" [ 2 4 0 1 3 0 0 0 149 0]\n",
" [ 1 3 0 1 11 0 1 0 1 104]]\n",
"val Loss: 0.1863 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[142 2 0 8 0 0 4 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 6 0 11 0 0]\n",
" [ 2 0 0 125 2 0 0 0 0 0]\n",
" [ 0 3 1 4 224 0 0 0 0 1]\n",
" [ 0 0 16 0 2 99 0 6 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 60/119\n",
"----------\n",
"train Loss: 0.2708 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[240 0 0 14 2 0 5 0 0 2]\n",
" [ 6 257 0 0 11 1 0 0 2 4]\n",
" [ 0 0 171 1 0 23 1 10 0 0]\n",
" [ 6 0 1 198 6 2 2 1 0 0]\n",
" [ 0 15 0 5 361 6 0 0 2 3]\n",
" [ 0 0 10 2 1 185 0 8 0 0]\n",
" [ 2 0 0 3 1 0 165 0 0 0]\n",
" [ 0 0 8 0 0 6 0 168 0 0]\n",
" [ 1 10 0 2 2 0 0 0 143 1]\n",
" [ 1 6 0 0 7 0 0 0 0 108]]\n",
"val Loss: 0.1840 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[141 1 0 10 0 0 4 0 1 0]\n",
" [ 0 157 0 0 6 0 0 0 3 1]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 2 2 0 0 0 90 0]\n",
" [ 1 3 1 1 1 0 0 0 0 65]]\n",
"\n",
"Epoch 61/119\n",
"----------\n",
"train Loss: 0.2604 Acc: 0.9117\n",
"Confusion Matrix:\n",
"[[242 3 0 10 1 0 6 1 0 0]\n",
" [ 5 254 0 2 13 1 0 1 3 2]\n",
" [ 1 0 169 1 2 23 0 10 0 0]\n",
" [ 14 2 0 192 2 0 6 0 0 0]\n",
" [ 2 17 0 3 363 3 1 0 1 2]\n",
" [ 0 0 8 0 2 183 0 12 0 1]\n",
" [ 4 3 0 1 0 0 163 0 0 0]\n",
" [ 0 0 3 0 1 4 0 174 0 0]\n",
" [ 1 3 0 1 2 0 0 0 152 0]\n",
" [ 3 1 1 0 4 0 0 0 1 112]]\n",
"val Loss: 0.1725 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[144 2 0 4 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 2 1]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 6 1 1 225 0 0 0 0 0]\n",
" [ 0 0 9 0 2 109 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 62/119\n",
"----------\n",
"train Loss: 0.2384 Acc: 0.9195\n",
"Confusion Matrix:\n",
"[[240 3 0 10 3 0 6 0 0 1]\n",
" [ 5 258 1 0 12 0 0 0 1 4]\n",
" [ 1 0 175 0 1 22 0 7 0 0]\n",
" [ 10 1 0 188 6 0 9 0 2 0]\n",
" [ 5 6 3 3 373 0 0 0 2 0]\n",
" [ 0 0 11 1 1 185 0 8 0 0]\n",
" [ 6 1 0 3 0 0 161 0 0 0]\n",
" [ 0 0 2 0 0 4 0 175 0 1]\n",
" [ 0 5 0 1 1 0 0 0 152 0]\n",
" [ 2 2 0 1 2 0 0 1 0 114]]\n",
"val Loss: 0.1806 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[142 1 0 8 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 3 1 3 225 0 0 0 0 1]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 2 0 105 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 1 3 1 1 1 0 0 0 0 65]]\n",
"\n",
"Epoch 63/119\n",
"----------\n",
"train Loss: 0.2636 Acc: 0.9126\n",
"Confusion Matrix:\n",
"[[239 1 0 12 1 0 8 0 0 2]\n",
" [ 2 259 0 0 13 0 0 1 1 5]\n",
" [ 2 0 172 0 0 21 1 10 0 0]\n",
" [ 5 1 1 202 5 0 2 0 0 0]\n",
" [ 1 15 2 1 365 1 1 0 2 4]\n",
" [ 0 0 13 1 0 183 0 8 0 1]\n",
" [ 5 2 0 3 0 0 161 0 0 0]\n",
" [ 1 0 2 0 0 7 0 172 0 0]\n",
" [ 1 8 0 0 1 1 0 0 148 0]\n",
" [ 1 4 0 3 9 0 0 0 0 105]]\n",
"val Loss: 0.1779 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 9 0 9 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 3 1 2 225 0 0 0 0 2]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 1 1 0 0 0 0 0 67]]\n",
"\n",
"Epoch 64/119\n",
"----------\n",
"train Loss: 0.2751 Acc: 0.9104\n",
"Confusion Matrix:\n",
"[[240 1 0 13 2 0 6 0 0 1]\n",
" [ 4 255 0 1 16 0 0 0 3 2]\n",
" [ 1 0 175 2 0 18 0 10 0 0]\n",
" [ 7 2 1 197 2 0 5 0 1 1]\n",
" [ 2 11 2 6 366 0 2 0 1 2]\n",
" [ 0 0 19 1 0 177 0 9 0 0]\n",
" [ 7 0 0 5 1 1 157 0 0 0]\n",
" [ 0 0 3 0 0 8 0 170 0 1]\n",
" [ 0 2 0 0 6 0 1 0 149 1]\n",
" [ 1 2 0 1 3 0 0 0 0 115]]\n",
"val Loss: 0.1838 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 7 0 3 110 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 2 2 0 0 0 90 0]\n",
" [ 0 3 0 1 2 0 2 0 0 64]]\n",
"\n",
"Epoch 65/119\n",
"----------\n",
"train Loss: 0.2554 Acc: 0.9145\n",
"Confusion Matrix:\n",
"[[243 4 0 9 2 1 4 0 0 0]\n",
" [ 3 253 1 1 12 1 1 0 3 6]\n",
" [ 1 0 176 1 0 16 0 11 0 1]\n",
" [ 12 1 0 191 7 0 5 0 0 0]\n",
" [ 1 13 1 2 367 1 1 0 2 4]\n",
" [ 0 0 11 0 0 189 0 6 0 0]\n",
" [ 5 0 0 6 0 0 158 2 0 0]\n",
" [ 0 1 2 1 0 5 0 173 0 0]\n",
" [ 2 2 0 0 4 0 2 0 149 0]\n",
" [ 1 4 0 0 5 0 1 0 0 111]]\n",
"val Loss: 0.1778 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[143 2 0 5 0 0 6 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 10 0 6 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 4 0 1 227 1 0 0 0 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 0 0 0 0 0 67]]\n",
"\n",
"Epoch 66/119\n",
"----------\n",
"train Loss: 0.2664 Acc: 0.9104\n",
"Confusion Matrix:\n",
"[[247 2 0 10 1 0 3 0 0 0]\n",
" [ 3 256 0 0 15 0 1 0 0 6]\n",
" [ 0 0 172 1 0 23 0 9 0 1]\n",
" [ 15 1 1 186 5 2 3 0 1 2]\n",
" [ 1 10 2 2 374 1 1 0 1 0]\n",
" [ 0 0 17 1 1 177 0 9 0 1]\n",
" [ 1 1 0 7 2 0 160 0 0 0]\n",
" [ 0 0 7 0 0 4 1 170 0 0]\n",
" [ 0 5 0 3 2 0 0 0 148 1]\n",
" [ 1 5 0 0 4 0 0 0 1 111]]\n",
"val Loss: 0.1842 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 3 0]\n",
" [ 0 0 105 0 0 6 0 12 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 7 1 2 222 0 0 0 0 1]\n",
" [ 0 0 15 0 2 101 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 67/119\n",
"----------\n",
"train Loss: 0.2408 Acc: 0.9186\n",
"Confusion Matrix:\n",
"[[238 4 0 12 2 0 7 0 0 0]\n",
" [ 1 265 0 2 10 0 1 0 2 0]\n",
" [ 0 1 180 0 1 17 0 7 0 0]\n",
" [ 12 1 0 197 2 0 2 0 0 2]\n",
" [ 3 13 2 7 364 1 0 0 1 1]\n",
" [ 0 0 15 0 2 181 0 8 0 0]\n",
" [ 3 2 0 4 1 0 160 1 0 0]\n",
" [ 0 0 2 0 0 6 0 173 0 1]\n",
" [ 0 5 0 0 4 0 0 0 150 0]\n",
" [ 1 3 0 0 6 0 0 0 1 111]]\n",
"val Loss: 0.1828 Acc: 0.9412\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 1 0 0 126 2 0 0 0 0 0]\n",
" [ 0 2 0 2 228 1 0 0 0 0]\n",
" [ 0 0 10 0 2 107 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 1 1 1 0 0 0 0 65]]\n",
"\n",
"Epoch 68/119\n",
"----------\n",
"train Loss: 0.2572 Acc: 0.9149\n",
"Confusion Matrix:\n",
"[[225 5 0 16 3 0 12 0 1 1]\n",
" [ 4 261 0 0 12 0 1 0 0 3]\n",
" [ 2 1 183 1 2 13 0 3 0 1]\n",
" [ 8 3 0 198 5 0 1 0 0 1]\n",
" [ 2 11 0 3 361 2 2 1 5 5]\n",
" [ 0 0 12 0 3 184 1 6 0 0]\n",
" [ 4 1 0 2 1 0 163 0 0 0]\n",
" [ 0 0 4 0 0 3 0 175 0 0]\n",
" [ 0 2 0 2 1 0 0 0 154 0]\n",
" [ 0 4 1 1 5 0 1 1 2 107]]\n",
"val Loss: 0.1767 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[145 1 0 6 0 0 5 0 0 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 5 1 5 222 0 0 0 0 0]\n",
" [ 0 0 14 0 2 104 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 2 4 0 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 69/119\n",
"----------\n",
"train Loss: 0.2598 Acc: 0.9158\n",
"Confusion Matrix:\n",
"[[243 1 1 10 2 0 4 1 0 1]\n",
" [ 6 252 0 2 14 0 1 1 1 4]\n",
" [ 1 0 182 1 2 16 0 3 0 1]\n",
" [ 9 1 0 198 4 0 1 1 0 2]\n",
" [ 0 13 2 9 366 0 0 0 1 1]\n",
" [ 0 0 9 0 2 185 0 10 0 0]\n",
" [ 5 2 0 3 0 0 160 1 0 0]\n",
" [ 0 0 6 0 0 3 1 171 0 1]\n",
" [ 0 6 0 1 4 1 0 0 147 0]\n",
" [ 0 3 0 2 6 1 0 0 1 109]]\n",
"val Loss: 0.1843 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[140 1 0 8 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 9 0 9 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 4 1 4 224 0 0 0 0 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 70/119\n",
"----------\n",
"train Loss: 0.2412 Acc: 0.9131\n",
"Confusion Matrix:\n",
"[[241 2 0 14 2 0 4 0 0 0]\n",
" [ 4 257 0 1 10 1 0 0 2 6]\n",
" [ 0 0 176 0 0 19 0 10 0 1]\n",
" [ 12 1 0 191 5 1 3 0 2 1]\n",
" [ 2 12 0 2 368 2 1 0 2 3]\n",
" [ 0 0 13 1 1 182 1 8 0 0]\n",
" [ 5 0 0 5 1 0 159 0 0 1]\n",
" [ 0 0 4 0 0 7 0 171 0 0]\n",
" [ 0 6 0 1 3 0 0 0 149 0]\n",
" [ 0 3 0 0 6 0 0 0 0 113]]\n",
"val Loss: 0.1940 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[141 1 0 10 0 0 4 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 9 0 9 0 0]\n",
" [ 0 0 0 128 1 0 0 0 0 0]\n",
" [ 0 4 1 4 223 0 0 0 1 0]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 2 5 1 2 0 0 0 0 0 62]]\n",
"\n",
"Epoch 71/119\n",
"----------\n",
"train Loss: 0.2369 Acc: 0.9231\n",
"Confusion Matrix:\n",
"[[239 0 1 16 3 0 4 0 0 0]\n",
" [ 1 267 0 0 10 0 0 0 1 2]\n",
" [ 0 0 182 0 1 14 0 9 0 0]\n",
" [ 13 0 0 192 5 0 4 1 0 1]\n",
" [ 0 6 1 4 370 1 3 1 1 5]\n",
" [ 0 0 13 1 2 180 0 9 0 1]\n",
" [ 6 0 0 5 1 0 158 0 0 1]\n",
" [ 0 0 2 0 0 5 0 175 0 0]\n",
" [ 0 5 0 0 4 0 0 0 150 0]\n",
" [ 2 0 0 1 2 0 1 0 0 116]]\n",
"val Loss: 0.1910 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[140 1 0 8 0 0 7 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 108 0 0 5 0 10 0 0]\n",
" [ 1 0 0 126 1 0 1 0 0 0]\n",
" [ 0 4 1 4 224 0 0 0 0 0]\n",
" [ 0 0 17 0 1 100 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 72/119\n",
"----------\n",
"train Loss: 0.2636 Acc: 0.9122\n",
"Confusion Matrix:\n",
"[[242 3 0 13 0 0 4 0 0 1]\n",
" [ 1 259 1 0 14 0 0 1 3 2]\n",
" [ 1 0 170 0 0 22 0 12 0 1]\n",
" [ 9 3 1 198 2 1 1 0 0 1]\n",
" [ 0 14 2 4 366 1 2 0 0 3]\n",
" [ 0 0 11 0 1 186 0 8 0 0]\n",
" [ 5 2 0 6 0 1 157 0 0 0]\n",
" [ 0 0 8 0 1 6 0 167 0 0]\n",
" [ 2 1 0 1 4 0 1 0 150 0]\n",
" [ 1 3 0 1 5 1 0 0 1 110]]\n",
"val Loss: 0.1810 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[142 3 0 7 0 0 4 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 5 1 1 223 0 0 0 1 2]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 0 0 0 0 0 68]]\n",
"\n",
"Epoch 73/119\n",
"----------\n",
"train Loss: 0.2582 Acc: 0.9149\n",
"Confusion Matrix:\n",
"[[238 4 0 14 2 0 5 0 0 0]\n",
" [ 0 265 0 0 8 1 2 0 1 4]\n",
" [ 1 1 174 1 1 23 0 5 0 0]\n",
" [ 10 1 1 193 5 0 4 0 1 1]\n",
" [ 2 13 3 5 362 0 1 1 4 1]\n",
" [ 0 0 10 1 0 187 0 8 0 0]\n",
" [ 6 0 1 4 0 0 160 0 0 0]\n",
" [ 0 1 4 1 1 7 0 168 0 0]\n",
" [ 2 1 0 0 6 0 0 0 149 1]\n",
" [ 0 2 0 1 3 0 0 1 0 115]]\n",
"val Loss: 0.1745 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 7 1 2 223 0 0 0 0 0]\n",
" [ 0 0 16 0 2 102 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 74/119\n",
"----------\n",
"train Loss: 0.2464 Acc: 0.9190\n",
"Confusion Matrix:\n",
"[[250 1 0 7 3 0 1 0 0 1]\n",
" [ 1 261 0 2 11 0 0 0 0 6]\n",
" [ 2 1 177 0 1 18 0 7 0 0]\n",
" [ 14 1 0 195 3 0 3 0 0 0]\n",
" [ 4 13 2 5 360 2 0 0 4 2]\n",
" [ 0 0 12 0 1 184 1 8 0 0]\n",
" [ 9 2 0 1 1 0 158 0 0 0]\n",
" [ 1 0 3 0 0 1 0 177 0 0]\n",
" [ 0 4 0 2 5 0 0 0 148 0]\n",
" [ 0 6 0 0 6 0 0 0 0 110]]\n",
"val Loss: 0.1881 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[142 1 0 6 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 3 1 4 225 0 0 0 0 0]\n",
" [ 0 0 15 0 2 101 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 1 4 0 1 0 0 2 0 0 64]]\n",
"\n",
"Epoch 75/119\n",
"----------\n",
"train Loss: 0.2515 Acc: 0.9108\n",
"Confusion Matrix:\n",
"[[240 3 1 11 1 0 3 0 0 4]\n",
" [ 1 256 0 0 14 0 0 1 3 6]\n",
" [ 0 0 169 0 0 21 1 15 0 0]\n",
" [ 4 2 1 201 7 0 1 0 0 0]\n",
" [ 3 15 2 6 364 1 1 0 0 0]\n",
" [ 0 0 12 1 0 183 0 10 0 0]\n",
" [ 3 1 0 5 1 0 161 0 0 0]\n",
" [ 0 0 7 0 1 7 0 167 0 0]\n",
" [ 1 5 0 0 3 0 0 0 149 1]\n",
" [ 1 4 0 1 3 0 1 0 0 112]]\n",
"val Loss: 0.1773 Acc: 0.9358\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 2 227 0 0 0 0 1]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 76/119\n",
"----------\n",
"train Loss: 0.2310 Acc: 0.9268\n",
"Confusion Matrix:\n",
"[[253 0 0 6 1 0 3 0 0 0]\n",
" [ 2 256 0 0 15 0 0 0 6 2]\n",
" [ 0 0 173 1 2 20 0 10 0 0]\n",
" [ 6 0 0 205 2 0 3 0 0 0]\n",
" [ 3 9 3 3 369 2 0 0 2 1]\n",
" [ 0 0 13 0 2 182 0 9 0 0]\n",
" [ 4 1 0 4 1 0 161 0 0 0]\n",
" [ 0 0 10 0 0 1 0 171 0 0]\n",
" [ 0 2 0 1 3 0 0 0 152 1]\n",
" [ 1 2 0 0 4 0 0 0 0 115]]\n",
"val Loss: 0.1774 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[144 1 0 7 0 0 4 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 3 1 4 225 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 77/119\n",
"----------\n",
"train Loss: 0.2614 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[238 1 0 17 1 0 6 0 0 0]\n",
" [ 6 255 2 2 13 0 0 0 1 2]\n",
" [ 0 2 178 0 0 20 0 5 0 1]\n",
" [ 10 2 0 199 3 0 2 0 0 0]\n",
" [ 2 12 2 6 362 2 2 0 0 4]\n",
" [ 0 1 14 0 2 178 0 11 0 0]\n",
" [ 0 0 0 4 4 0 163 0 0 0]\n",
" [ 0 1 7 1 0 6 0 167 0 0]\n",
" [ 0 3 0 2 7 0 0 0 147 0]\n",
" [ 0 5 0 2 2 0 1 1 2 109]]\n",
"val Loss: 0.1787 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[144 1 0 7 0 0 4 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 11 0 5 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 3 1 5 222 0 0 0 0 2]\n",
" [ 0 0 16 0 2 102 0 3 0 0]\n",
" [ 1 0 0 0 0 0 100 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 2 0 1 0 0 0 0 0 69]]\n",
"\n",
"Epoch 78/119\n",
"----------\n",
"train Loss: 0.2707 Acc: 0.9049\n",
"Confusion Matrix:\n",
"[[240 4 0 12 2 0 5 0 0 0]\n",
" [ 9 258 1 0 12 0 1 0 0 0]\n",
" [ 1 3 169 2 2 22 0 6 0 1]\n",
" [ 12 1 0 197 4 0 2 0 0 0]\n",
" [ 1 14 1 3 365 0 2 0 4 2]\n",
" [ 1 1 19 0 3 172 0 10 0 0]\n",
" [ 5 0 0 3 1 0 162 0 0 0]\n",
" [ 0 0 5 0 0 10 0 167 0 0]\n",
" [ 1 8 0 0 3 0 0 0 147 0]\n",
" [ 3 1 0 2 4 0 0 0 0 112]]\n",
"val Loss: 0.1799 Acc: 0.9305\n",
"Confusion Matrix:\n",
"[[144 1 0 5 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 9 0 8 0 0]\n",
" [ 4 0 0 121 2 0 2 0 0 0]\n",
" [ 0 5 1 2 222 1 0 0 0 2]\n",
" [ 0 0 18 0 1 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 1 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 79/119\n",
"----------\n",
"train Loss: 0.2710 Acc: 0.9035\n",
"Confusion Matrix:\n",
"[[242 0 1 13 3 0 2 0 1 1]\n",
" [ 1 257 2 1 15 0 1 0 1 3]\n",
" [ 0 0 164 1 3 25 1 10 0 2]\n",
" [ 11 1 0 195 5 0 4 0 0 0]\n",
" [ 5 12 1 3 366 3 1 0 0 1]\n",
" [ 0 0 20 0 1 173 0 12 0 0]\n",
" [ 8 1 0 5 2 0 155 0 0 0]\n",
" [ 0 0 3 0 2 2 0 175 0 0]\n",
" [ 0 5 1 0 3 0 1 0 149 0]\n",
" [ 1 6 0 1 3 0 0 1 0 110]]\n",
"val Loss: 0.1832 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 3 1 2 227 0 0 0 0 0]\n",
" [ 0 0 17 0 2 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 80/119\n",
"----------\n",
"train Loss: 0.2557 Acc: 0.9163\n",
"Confusion Matrix:\n",
"[[238 5 0 13 2 0 5 0 0 0]\n",
" [ 2 261 1 1 12 1 0 0 2 1]\n",
" [ 0 1 172 0 1 19 0 10 0 3]\n",
" [ 7 3 0 199 4 0 2 0 1 0]\n",
" [ 0 4 1 2 379 0 0 0 2 4]\n",
" [ 0 0 20 1 0 171 0 14 0 0]\n",
" [ 2 1 0 1 0 0 167 0 0 0]\n",
" [ 0 0 6 0 0 8 0 168 0 0]\n",
" [ 1 3 0 0 6 0 0 0 149 0]\n",
" [ 1 3 1 0 6 0 1 0 0 110]]\n",
"val Loss: 0.1769 Acc: 0.9404\n",
"Confusion Matrix:\n",
"[[142 2 0 8 0 0 4 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 11 0 2 106 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 81/119\n",
"----------\n",
"train Loss: 0.2626 Acc: 0.9113\n",
"Confusion Matrix:\n",
"[[241 2 1 10 1 0 7 0 0 1]\n",
" [ 2 263 0 1 6 0 2 1 3 3]\n",
" [ 0 0 165 1 1 22 1 13 0 3]\n",
" [ 7 1 0 196 4 0 5 0 1 2]\n",
" [ 1 10 0 2 370 1 1 0 4 3]\n",
" [ 0 0 11 0 1 186 0 8 0 0]\n",
" [ 4 1 1 7 1 0 157 0 0 0]\n",
" [ 1 0 7 0 0 9 0 164 0 1]\n",
" [ 0 3 0 0 3 0 1 0 152 0]\n",
" [ 0 5 1 1 4 0 1 1 0 109]]\n",
"val Loss: 0.1819 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[143 2 0 6 0 0 5 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 106 0 0 6 0 11 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 8 1 2 221 0 0 0 1 0]\n",
" [ 0 0 10 0 2 106 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 0 0 0 66]]\n",
"\n",
"Epoch 82/119\n",
"----------\n",
"train Loss: 0.2514 Acc: 0.9158\n",
"Confusion Matrix:\n",
"[[240 3 0 13 4 0 2 0 1 0]\n",
" [ 1 261 0 2 11 1 0 0 1 4]\n",
" [ 1 0 170 0 0 22 0 11 0 2]\n",
" [ 7 2 0 200 4 0 3 0 0 0]\n",
" [ 3 11 2 4 365 1 1 0 1 4]\n",
" [ 0 0 16 0 0 182 0 8 0 0]\n",
" [ 3 2 0 2 1 0 162 1 0 0]\n",
" [ 0 0 6 0 0 5 0 171 0 0]\n",
" [ 0 4 0 2 4 0 0 0 149 0]\n",
" [ 0 3 0 2 3 0 1 0 0 113]]\n",
"val Loss: 0.1781 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 6 1 2 222 0 0 0 0 2]\n",
" [ 0 0 13 0 2 105 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 3 0 0 3 0 103 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 83/119\n",
"----------\n",
"train Loss: 0.2459 Acc: 0.9177\n",
"Confusion Matrix:\n",
"[[233 4 0 14 3 0 6 1 0 2]\n",
" [ 2 265 1 0 10 0 0 0 0 3]\n",
" [ 1 0 177 0 1 17 0 8 0 2]\n",
" [ 5 2 0 204 1 0 3 0 0 1]\n",
" [ 3 7 1 6 368 4 1 0 1 1]\n",
" [ 1 1 18 0 0 177 0 9 0 0]\n",
" [ 2 1 0 1 1 0 166 0 0 0]\n",
" [ 0 0 9 1 0 7 0 165 0 0]\n",
" [ 1 6 0 1 2 0 0 0 149 0]\n",
" [ 1 1 0 0 6 0 0 1 0 113]]\n",
"val Loss: 0.1839 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 1 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 84/119\n",
"----------\n",
"train Loss: 0.2371 Acc: 0.9181\n",
"Confusion Matrix:\n",
"[[242 2 0 13 3 0 3 0 0 0]\n",
" [ 2 265 0 0 12 0 1 0 1 0]\n",
" [ 0 1 178 0 2 14 1 9 0 1]\n",
" [ 20 0 0 193 0 0 2 0 1 0]\n",
" [ 0 10 1 3 371 2 3 0 0 2]\n",
" [ 0 0 14 0 1 177 0 13 0 1]\n",
" [ 4 0 0 2 0 0 165 0 0 0]\n",
" [ 0 0 6 1 0 10 0 165 0 0]\n",
" [ 0 3 0 1 2 0 0 0 153 0]\n",
" [ 1 5 0 2 4 0 0 0 1 109]]\n",
"val Loss: 0.1837 Acc: 0.9404\n",
"Confusion Matrix:\n",
"[[143 1 0 7 0 0 5 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 6 0 11 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 1 0 1 0 0 66]]\n",
"\n",
"Epoch 85/119\n",
"----------\n",
"train Loss: 0.2606 Acc: 0.9108\n",
"Confusion Matrix:\n",
"[[238 4 1 11 1 0 6 0 1 1]\n",
" [ 5 256 0 1 11 0 0 0 2 6]\n",
" [ 1 1 172 1 1 22 1 6 0 1]\n",
" [ 11 3 1 195 2 0 4 0 0 0]\n",
" [ 0 17 2 1 367 1 2 0 1 1]\n",
" [ 0 0 15 0 2 180 0 9 0 0]\n",
" [ 2 1 0 5 1 0 162 0 0 0]\n",
" [ 0 1 3 0 1 3 0 174 0 0]\n",
" [ 1 5 0 1 4 0 0 0 148 0]\n",
" [ 2 3 0 1 6 0 0 0 0 110]]\n",
"val Loss: 0.1830 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 3 1 4 225 0 0 0 0 0]\n",
" [ 0 0 17 0 2 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 86/119\n",
"----------\n",
"train Loss: 0.2546 Acc: 0.9154\n",
"Confusion Matrix:\n",
"[[239 3 0 15 4 0 2 0 0 0]\n",
" [ 1 268 0 1 9 0 0 0 1 1]\n",
" [ 2 0 175 0 0 20 1 8 0 0]\n",
" [ 8 2 0 193 3 1 7 1 0 1]\n",
" [ 2 15 0 2 369 3 0 0 0 1]\n",
" [ 0 0 17 1 2 180 0 6 0 0]\n",
" [ 7 1 0 3 1 0 157 0 0 2]\n",
" [ 1 0 3 0 0 5 0 173 0 0]\n",
" [ 1 3 0 1 2 0 1 0 151 0]\n",
" [ 1 2 1 2 6 1 0 0 2 107]]\n",
"val Loss: 0.1825 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[140 2 0 8 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 17 0 2 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 0 0 0 0 0 67]]\n",
"\n",
"Epoch 87/119\n",
"----------\n",
"train Loss: 0.2704 Acc: 0.9081\n",
"Confusion Matrix:\n",
"[[248 0 1 7 0 0 6 0 1 0]\n",
" [ 4 257 1 0 12 0 0 1 2 4]\n",
" [ 0 2 169 1 0 20 0 13 0 1]\n",
" [ 8 2 1 195 5 0 3 1 0 1]\n",
" [ 2 8 0 4 369 2 1 0 5 1]\n",
" [ 0 0 19 0 5 172 0 10 0 0]\n",
" [ 4 0 0 5 2 0 160 0 0 0]\n",
" [ 0 0 6 0 0 6 0 169 0 1]\n",
" [ 0 4 0 2 2 0 0 0 151 0]\n",
" [ 2 6 2 1 5 0 0 0 0 106]]\n",
"val Loss: 0.1803 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[142 2 0 6 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 3 1 2 224 1 0 0 0 2]\n",
" [ 0 0 16 0 1 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 2 0 105 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 2 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 88/119\n",
"----------\n",
"train Loss: 0.2437 Acc: 0.9181\n",
"Confusion Matrix:\n",
"[[234 3 0 14 3 0 6 1 0 2]\n",
" [ 1 263 0 1 12 0 1 0 2 1]\n",
" [ 0 0 182 0 0 17 0 7 0 0]\n",
" [ 13 2 0 197 3 0 0 1 0 0]\n",
" [ 2 10 1 1 371 2 1 0 1 3]\n",
" [ 0 0 16 0 1 181 0 7 0 1]\n",
" [ 5 0 0 6 1 0 159 0 0 0]\n",
" [ 0 0 3 0 1 5 1 172 0 0]\n",
" [ 0 5 0 2 5 0 0 0 146 1]\n",
" [ 1 2 1 1 1 0 1 1 1 113]]\n",
"val Loss: 0.1871 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 2 0 0 125 1 0 1 0 0 0]\n",
" [ 0 4 1 5 222 0 0 0 1 0]\n",
" [ 0 0 17 0 1 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 2 4 0 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 89/119\n",
"----------\n",
"train Loss: 0.2443 Acc: 0.9140\n",
"Confusion Matrix:\n",
"[[232 3 0 20 1 1 5 0 0 1]\n",
" [ 2 263 0 0 10 0 1 0 1 4]\n",
" [ 0 0 178 0 0 18 0 7 0 3]\n",
" [ 7 2 0 198 5 1 2 0 0 1]\n",
" [ 2 6 0 4 371 4 2 0 1 2]\n",
" [ 0 0 18 0 3 172 0 12 0 1]\n",
" [ 4 0 0 4 2 0 161 0 0 0]\n",
" [ 0 0 2 0 0 2 0 178 0 0]\n",
" [ 0 2 0 1 7 0 0 0 149 0]\n",
" [ 1 6 1 0 4 0 2 0 1 107]]\n",
"val Loss: 0.1878 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[141 1 0 8 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 5 0 11 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 5 224 0 0 0 0 1]\n",
" [ 0 0 18 0 1 99 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 2 1 1 1 0 1 0 0 66]]\n",
"\n",
"Epoch 90/119\n",
"----------\n",
"train Loss: 0.2478 Acc: 0.9186\n",
"Confusion Matrix:\n",
"[[242 0 0 12 3 0 5 0 1 0]\n",
" [ 5 262 1 1 7 0 0 1 1 3]\n",
" [ 2 1 173 0 0 19 0 10 0 1]\n",
" [ 9 2 0 197 4 0 3 0 0 1]\n",
" [ 0 7 3 3 374 2 1 0 1 1]\n",
" [ 0 0 21 0 0 177 1 7 0 0]\n",
" [ 2 1 0 2 1 1 164 0 0 0]\n",
" [ 0 1 3 0 0 7 0 171 0 0]\n",
" [ 0 6 0 3 2 0 0 0 148 0]\n",
" [ 0 2 2 0 7 0 0 0 0 111]]\n",
"val Loss: 0.1818 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[140 1 0 9 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 4 225 0 0 0 0 1]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 3 0 1 1 0 2 0 0 65]]\n",
"\n",
"Epoch 91/119\n",
"----------\n",
"train Loss: 0.2245 Acc: 0.9290\n",
"Confusion Matrix:\n",
"[[244 1 0 13 0 0 2 0 1 2]\n",
" [ 2 267 0 0 10 0 0 0 1 1]\n",
" [ 1 0 177 0 0 18 0 10 0 0]\n",
" [ 7 1 1 197 3 0 2 0 1 4]\n",
" [ 0 8 1 3 375 2 0 0 0 3]\n",
" [ 0 1 9 0 1 188 0 7 0 0]\n",
" [ 7 1 0 4 1 0 158 0 0 0]\n",
" [ 0 0 3 0 0 9 0 170 0 0]\n",
" [ 0 3 0 0 1 0 0 0 155 0]\n",
" [ 1 2 1 2 5 0 0 0 0 111]]\n",
"val Loss: 0.1821 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 3 1 2 225 0 0 0 0 2]\n",
" [ 0 0 14 0 2 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 92/119\n",
"----------\n",
"train Loss: 0.2459 Acc: 0.9195\n",
"Confusion Matrix:\n",
"[[240 0 0 16 1 0 5 0 0 1]\n",
" [ 4 260 0 0 11 0 1 0 0 5]\n",
" [ 0 0 183 1 0 15 0 7 0 0]\n",
" [ 9 1 0 199 4 0 2 0 1 0]\n",
" [ 2 16 1 6 359 1 0 0 5 2]\n",
" [ 0 0 10 0 1 185 0 9 0 1]\n",
" [ 4 1 0 4 1 0 161 0 0 0]\n",
" [ 0 0 3 0 0 4 0 175 0 0]\n",
" [ 0 5 0 0 7 0 0 0 147 0]\n",
" [ 0 4 1 2 2 0 1 0 0 112]]\n",
"val Loss: 0.1808 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 104 0 0 8 0 11 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 5 1 3 224 0 0 0 0 0]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 0 0 108 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 93/119\n",
"----------\n",
"train Loss: 0.2399 Acc: 0.9177\n",
"Confusion Matrix:\n",
"[[244 3 0 12 0 0 4 0 0 0]\n",
" [ 3 263 0 0 11 0 0 0 2 2]\n",
" [ 2 0 173 0 1 18 0 9 0 3]\n",
" [ 12 3 0 193 4 0 4 0 0 0]\n",
" [ 2 5 2 0 377 1 2 0 1 2]\n",
" [ 1 1 12 1 3 178 0 10 0 0]\n",
" [ 5 2 0 4 3 0 157 0 0 0]\n",
" [ 0 0 7 0 0 7 0 168 0 0]\n",
" [ 1 2 0 1 4 0 0 0 151 0]\n",
" [ 0 2 0 2 4 0 1 0 0 113]]\n",
"val Loss: 0.1807 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 2 0 0 0 0 66]]\n",
"\n",
"Epoch 94/119\n",
"----------\n",
"train Loss: 0.2565 Acc: 0.9131\n",
"Confusion Matrix:\n",
"[[240 0 0 16 0 0 6 0 0 1]\n",
" [ 3 268 0 0 6 0 0 0 1 3]\n",
" [ 0 0 179 0 0 20 0 7 0 0]\n",
" [ 10 1 1 194 6 0 3 0 1 0]\n",
" [ 3 12 2 3 364 2 2 0 1 3]\n",
" [ 0 1 18 0 2 173 0 11 0 1]\n",
" [ 8 1 0 1 2 0 159 0 0 0]\n",
" [ 0 0 6 0 0 2 0 174 0 0]\n",
" [ 0 3 0 2 3 0 0 0 150 1]\n",
" [ 1 4 1 0 9 0 0 0 1 106]]\n",
"val Loss: 0.1805 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[142 2 0 7 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 2 0 0 125 2 0 0 0 0 0]\n",
" [ 0 5 1 2 225 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 95/119\n",
"----------\n",
"train Loss: 0.2623 Acc: 0.9104\n",
"Confusion Matrix:\n",
"[[239 3 1 15 2 0 1 2 0 0]\n",
" [ 4 259 0 1 10 0 0 0 3 4]\n",
" [ 3 0 175 0 0 15 1 9 0 3]\n",
" [ 11 2 0 194 3 1 3 1 1 0]\n",
" [ 2 10 1 6 364 1 2 0 3 3]\n",
" [ 0 1 10 0 1 182 0 11 0 1]\n",
" [ 8 1 0 5 0 0 157 0 0 0]\n",
" [ 0 0 2 0 0 4 0 176 0 0]\n",
" [ 0 11 0 3 4 0 0 0 141 0]\n",
" [ 2 3 0 1 2 0 0 0 0 114]]\n",
"val Loss: 0.1786 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[144 1 0 5 0 0 6 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 109 0 0 9 0 5 0 0]\n",
" [ 4 0 0 121 2 0 2 0 0 0]\n",
" [ 0 7 1 3 222 0 0 0 0 0]\n",
" [ 0 0 16 0 2 102 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 2 0 105 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 5 0 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 96/119\n",
"----------\n",
"train Loss: 0.2496 Acc: 0.9145\n",
"Confusion Matrix:\n",
"[[237 3 0 15 1 0 7 0 0 0]\n",
" [ 3 263 0 1 11 0 0 1 1 1]\n",
" [ 0 0 183 1 1 11 0 10 0 0]\n",
" [ 11 3 1 193 1 0 4 0 2 1]\n",
" [ 1 17 1 5 362 2 1 0 1 2]\n",
" [ 0 0 15 0 4 179 0 8 0 0]\n",
" [ 7 0 0 4 0 0 160 0 0 0]\n",
" [ 0 0 6 0 0 6 0 170 0 0]\n",
" [ 0 4 0 1 3 0 1 0 150 0]\n",
" [ 1 1 0 2 4 0 0 0 1 113]]\n",
"val Loss: 0.1843 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 9 0 9 0 0]\n",
" [ 2 0 0 123 2 0 2 0 0 0]\n",
" [ 0 2 1 4 226 0 0 0 0 0]\n",
" [ 0 0 14 0 2 104 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 3 0 1 1 0 2 0 0 65]]\n",
"\n",
"Epoch 97/119\n",
"----------\n",
"train Loss: 0.2436 Acc: 0.9217\n",
"Confusion Matrix:\n",
"[[250 0 0 11 0 0 2 0 0 0]\n",
" [ 4 252 1 0 17 0 0 0 2 5]\n",
" [ 0 0 176 1 1 18 0 9 0 1]\n",
" [ 4 4 0 198 3 0 5 0 2 0]\n",
" [ 1 12 0 5 369 1 1 0 2 1]\n",
" [ 0 0 12 0 0 185 0 9 0 0]\n",
" [ 2 0 0 3 1 0 165 0 0 0]\n",
" [ 0 0 7 0 0 7 0 168 0 0]\n",
" [ 0 4 0 1 2 0 0 0 151 1]\n",
" [ 0 3 0 0 5 0 1 0 1 112]]\n",
"val Loss: 0.1792 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[144 1 0 5 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 5 1 3 224 0 0 0 0 0]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 98/119\n",
"----------\n",
"train Loss: 0.2723 Acc: 0.9076\n",
"Confusion Matrix:\n",
"[[241 4 1 12 0 0 3 0 0 2]\n",
" [ 4 254 0 1 13 0 1 1 5 2]\n",
" [ 1 1 178 0 2 18 0 6 0 0]\n",
" [ 9 1 0 194 6 1 4 0 0 1]\n",
" [ 2 16 0 4 363 3 2 2 0 0]\n",
" [ 0 0 22 0 1 173 0 10 0 0]\n",
" [ 2 0 1 5 1 0 161 0 0 1]\n",
" [ 0 0 6 0 0 8 0 168 0 0]\n",
" [ 0 6 0 1 2 0 0 0 150 0]\n",
" [ 0 3 0 1 4 0 1 0 0 113]]\n",
"val Loss: 0.1821 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[141 1 0 7 0 0 7 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 2 1 2 228 0 0 0 0 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 1 0 106 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 3 1 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 99/119\n",
"----------\n",
"train Loss: 0.2522 Acc: 0.9163\n",
"Confusion Matrix:\n",
"[[239 2 0 17 0 0 5 0 0 0]\n",
" [ 3 251 0 1 16 0 1 1 2 6]\n",
" [ 1 0 179 1 0 16 0 9 0 0]\n",
" [ 13 1 0 193 5 0 4 0 0 0]\n",
" [ 2 15 1 3 367 0 0 0 1 3]\n",
" [ 0 0 8 0 1 181 0 16 0 0]\n",
" [ 3 0 0 2 0 0 164 1 0 1]\n",
" [ 0 0 5 0 0 3 0 174 0 0]\n",
" [ 2 2 0 1 5 0 0 0 149 0]\n",
" [ 1 0 0 0 3 0 0 0 1 117]]\n",
"val Loss: 0.1804 Acc: 0.9297\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 107 0 0 10 0 6 0 0]\n",
" [ 3 0 0 121 2 0 3 0 0 0]\n",
" [ 0 6 1 4 222 0 0 0 0 0]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 2 4 0 1 0 0 1 0 0 64]]\n",
"\n",
"Epoch 100/119\n",
"----------\n",
"train Loss: 0.2452 Acc: 0.9240\n",
"Confusion Matrix:\n",
"[[240 2 1 16 1 0 2 0 0 1]\n",
" [ 2 263 0 1 9 0 0 0 1 5]\n",
" [ 1 0 178 0 2 19 0 6 0 0]\n",
" [ 12 0 0 193 8 0 3 0 0 0]\n",
" [ 1 16 2 2 366 0 1 1 3 0]\n",
" [ 0 1 8 0 1 191 0 5 0 0]\n",
" [ 5 0 0 4 1 0 161 0 0 0]\n",
" [ 0 0 5 1 0 6 0 170 0 0]\n",
" [ 1 1 0 0 2 0 0 0 155 0]\n",
" [ 1 3 0 1 3 0 0 0 0 114]]\n",
"val Loss: 0.1869 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[141 1 0 7 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 1 0 0 126 1 0 1 0 0 0]\n",
" [ 0 3 1 4 225 0 0 0 0 0]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 3 0 1 1 0 2 0 0 65]]\n",
"\n",
"Epoch 101/119\n",
"----------\n",
"train Loss: 0.2533 Acc: 0.9136\n",
"Confusion Matrix:\n",
"[[245 2 2 11 1 0 2 0 0 0]\n",
" [ 2 257 0 0 17 0 0 0 2 3]\n",
" [ 1 1 176 1 1 20 0 5 0 1]\n",
" [ 12 3 0 190 6 0 5 0 0 0]\n",
" [ 3 8 3 8 362 3 1 0 0 4]\n",
" [ 0 0 15 1 0 187 0 3 0 0]\n",
" [ 2 0 0 3 3 0 163 0 0 0]\n",
" [ 1 0 2 0 0 9 1 169 0 0]\n",
" [ 0 2 0 0 8 1 0 0 148 0]\n",
" [ 0 5 0 0 4 0 1 0 1 111]]\n",
"val Loss: 0.1795 Acc: 0.9335\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 162 0 0 2 0 0 0 3 0]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 7 1 2 223 0 0 0 0 0]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 102/119\n",
"----------\n",
"train Loss: 0.2858 Acc: 0.9054\n",
"Confusion Matrix:\n",
"[[237 2 0 16 1 0 6 0 0 1]\n",
" [ 4 255 0 0 14 0 0 0 3 5]\n",
" [ 0 0 172 0 2 26 0 5 0 1]\n",
" [ 13 1 0 194 3 1 2 0 1 1]\n",
" [ 2 8 2 2 372 2 0 0 3 1]\n",
" [ 0 0 17 0 4 176 0 8 0 1]\n",
" [ 3 0 0 5 2 1 160 0 0 0]\n",
" [ 1 0 5 0 0 10 0 166 0 0]\n",
" [ 0 5 1 2 1 0 0 0 150 0]\n",
" [ 1 5 1 3 4 0 0 0 0 108]]\n",
"val Loss: 0.1803 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[144 1 0 6 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 7 0 9 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 3 225 0 0 0 0 2]\n",
" [ 0 0 16 0 2 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 2 0 1 1 0 0 0 0 67]]\n",
"\n",
"Epoch 103/119\n",
"----------\n",
"train Loss: 0.2413 Acc: 0.9154\n",
"Confusion Matrix:\n",
"[[239 4 1 13 1 0 5 0 0 0]\n",
" [ 2 261 0 2 9 0 1 0 3 3]\n",
" [ 1 1 166 0 2 28 0 7 0 1]\n",
" [ 8 3 0 195 2 0 6 2 0 0]\n",
" [ 0 6 1 3 376 0 2 0 2 2]\n",
" [ 1 0 16 0 0 181 0 8 0 0]\n",
" [ 2 0 1 2 0 0 166 0 0 0]\n",
" [ 0 1 5 0 0 6 0 170 0 0]\n",
" [ 0 7 0 0 5 0 0 0 146 1]\n",
" [ 1 5 0 0 2 0 2 0 0 112]]\n",
"val Loss: 0.1832 Acc: 0.9396\n",
"Confusion Matrix:\n",
"[[143 1 0 8 0 0 4 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 11 0 7 0 0]\n",
" [ 1 0 0 126 2 0 0 0 0 0]\n",
" [ 0 2 0 4 226 1 0 0 0 0]\n",
" [ 0 0 11 0 2 107 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 2 0 105 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 1 3 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 104/119\n",
"----------\n",
"train Loss: 0.2633 Acc: 0.9154\n",
"Confusion Matrix:\n",
"[[235 3 0 12 5 0 5 0 0 3]\n",
" [ 4 260 0 0 12 2 0 0 1 2]\n",
" [ 1 0 181 0 0 18 0 5 0 1]\n",
" [ 9 2 0 200 1 0 1 1 0 2]\n",
" [ 1 15 0 1 368 2 1 1 2 1]\n",
" [ 0 0 15 0 3 176 0 12 0 0]\n",
" [ 8 0 0 3 0 0 159 0 0 1]\n",
" [ 0 0 2 0 0 5 0 174 0 1]\n",
" [ 1 0 0 1 10 0 0 0 147 0]\n",
" [ 0 1 0 0 7 1 0 0 1 112]]\n",
"val Loss: 0.1878 Acc: 0.9343\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 123 2 0 2 0 0 0]\n",
" [ 0 4 1 3 223 0 0 0 1 1]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 105/119\n",
"----------\n",
"train Loss: 0.2516 Acc: 0.9122\n",
"Confusion Matrix:\n",
"[[243 1 1 10 1 0 7 0 0 0]\n",
" [ 3 259 0 2 12 1 0 0 1 3]\n",
" [ 0 0 172 1 1 26 1 5 0 0]\n",
" [ 14 1 0 187 4 0 9 0 1 0]\n",
" [ 1 10 1 3 371 3 0 1 0 2]\n",
" [ 0 0 16 0 1 176 0 13 0 0]\n",
" [ 3 0 0 6 1 0 161 0 0 0]\n",
" [ 0 0 5 0 0 7 0 170 0 0]\n",
" [ 0 2 0 0 7 0 0 0 150 0]\n",
" [ 1 0 0 0 5 0 0 0 0 116]]\n",
"val Loss: 0.1799 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[141 2 0 6 0 0 7 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 2 0 1 2 0 0 0 0 67]]\n",
"\n",
"Epoch 106/119\n",
"----------\n",
"train Loss: 0.2277 Acc: 0.9304\n",
"Confusion Matrix:\n",
"[[247 0 0 10 1 0 3 0 1 1]\n",
" [ 2 264 0 0 10 0 2 1 2 0]\n",
" [ 0 0 177 0 1 18 0 9 0 1]\n",
" [ 11 1 0 194 6 0 4 0 0 0]\n",
" [ 1 9 0 1 372 1 0 1 3 4]\n",
" [ 0 0 7 0 2 195 0 2 0 0]\n",
" [ 7 0 0 5 0 0 159 0 0 0]\n",
" [ 0 0 2 0 0 5 0 175 0 0]\n",
" [ 0 3 0 2 3 0 0 0 151 0]\n",
" [ 1 5 0 0 4 0 1 0 0 111]]\n",
"val Loss: 0.1863 Acc: 0.9389\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 107 0 0 9 0 7 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 3 1 2 226 0 0 0 1 0]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 4 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 107/119\n",
"----------\n",
"train Loss: 0.2626 Acc: 0.9167\n",
"Confusion Matrix:\n",
"[[247 1 0 10 2 0 2 0 0 1]\n",
" [ 5 260 0 0 8 0 1 1 2 4]\n",
" [ 0 0 172 0 1 19 2 12 0 0]\n",
" [ 11 1 0 196 4 0 4 0 0 0]\n",
" [ 3 15 0 6 363 2 0 0 2 1]\n",
" [ 1 1 12 0 2 181 0 9 0 0]\n",
" [ 4 0 0 5 0 0 162 0 0 0]\n",
" [ 0 0 7 1 0 4 0 169 0 1]\n",
" [ 0 3 1 1 2 0 0 0 152 0]\n",
" [ 1 3 0 0 4 0 0 0 1 113]]\n",
"val Loss: 0.1836 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 8 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 3 1 3 224 0 0 0 0 2]\n",
" [ 0 0 17 0 2 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 108/119\n",
"----------\n",
"train Loss: 0.2472 Acc: 0.9208\n",
"Confusion Matrix:\n",
"[[240 4 0 10 2 0 5 1 1 0]\n",
" [ 2 263 0 0 14 0 0 0 1 1]\n",
" [ 0 0 178 0 0 20 0 8 0 0]\n",
" [ 7 2 0 201 4 1 0 1 0 0]\n",
" [ 2 5 1 2 373 0 1 0 3 5]\n",
" [ 0 0 10 0 2 179 0 15 0 0]\n",
" [ 3 0 0 4 1 0 163 0 0 0]\n",
" [ 0 0 9 0 1 5 0 166 0 1]\n",
" [ 0 2 0 1 4 0 0 0 152 0]\n",
" [ 0 5 2 0 5 0 1 0 0 109]]\n",
"val Loss: 0.1807 Acc: 0.9404\n",
"Confusion Matrix:\n",
"[[143 1 0 7 0 0 5 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 2 1 3 227 0 0 0 0 0]\n",
" [ 0 0 12 0 2 105 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 1 3 0 1 1 0 0 0 0 66]]\n",
"\n",
"Epoch 109/119\n",
"----------\n",
"train Loss: 0.2602 Acc: 0.9086\n",
"Confusion Matrix:\n",
"[[235 3 1 18 0 0 5 0 0 1]\n",
" [ 4 256 0 1 11 1 1 1 4 2]\n",
" [ 2 0 172 1 2 19 1 9 0 0]\n",
" [ 13 2 0 191 3 0 7 0 0 0]\n",
" [ 0 12 1 2 370 1 2 0 1 3]\n",
" [ 0 0 10 0 2 183 0 11 0 0]\n",
" [ 4 1 0 4 3 0 158 1 0 0]\n",
" [ 0 0 8 1 1 5 0 167 0 0]\n",
" [ 0 2 0 0 2 0 0 0 154 1]\n",
" [ 0 2 1 1 6 0 1 0 0 111]]\n",
"val Loss: 0.1940 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[140 1 0 8 0 0 7 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 107 0 0 8 0 8 0 0]\n",
" [ 1 0 0 125 1 0 2 0 0 0]\n",
" [ 0 3 1 5 222 0 0 0 0 2]\n",
" [ 0 0 15 0 1 103 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 1 4 0 1 0 0 2 0 0 64]]\n",
"\n",
"Epoch 110/119\n",
"----------\n",
"train Loss: 0.2367 Acc: 0.9172\n",
"Confusion Matrix:\n",
"[[240 4 0 9 2 0 6 0 0 2]\n",
" [ 2 245 0 1 28 0 0 0 2 3]\n",
" [ 1 0 185 0 1 14 0 5 0 0]\n",
" [ 15 2 0 195 3 0 1 0 0 0]\n",
" [ 1 14 1 3 372 0 0 0 0 1]\n",
" [ 0 0 10 1 1 182 0 12 0 0]\n",
" [ 8 0 0 1 1 0 161 0 0 0]\n",
" [ 0 0 1 0 0 11 0 170 0 0]\n",
" [ 0 2 0 1 6 0 0 0 149 1]\n",
" [ 0 2 0 0 2 0 1 0 0 117]]\n",
"val Loss: 0.1827 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[143 1 0 5 0 0 7 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 8 0 9 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 2 1 3 226 0 0 0 0 1]\n",
" [ 0 0 18 0 1 100 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 0 0 0 0 92 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 111/119\n",
"----------\n",
"train Loss: 0.2503 Acc: 0.9108\n",
"Confusion Matrix:\n",
"[[239 2 0 15 2 0 4 0 0 1]\n",
" [ 0 261 0 0 15 0 0 0 2 3]\n",
" [ 0 0 177 0 0 21 0 8 0 0]\n",
" [ 8 1 0 198 5 0 4 0 0 0]\n",
" [ 2 9 0 2 368 1 1 1 4 4]\n",
" [ 0 0 15 0 1 180 0 10 0 0]\n",
" [ 4 0 0 5 0 0 160 1 0 1]\n",
" [ 1 0 8 0 0 4 0 169 0 0]\n",
" [ 1 6 0 1 4 0 0 0 147 0]\n",
" [ 2 5 0 2 7 1 1 1 0 103]]\n",
"val Loss: 0.1861 Acc: 0.9328\n",
"Confusion Matrix:\n",
"[[143 2 0 6 0 0 5 0 1 0]\n",
" [ 0 161 0 0 3 0 0 0 3 0]\n",
" [ 0 0 105 0 0 6 0 12 0 0]\n",
" [ 4 0 0 122 2 0 1 0 0 0]\n",
" [ 0 8 1 2 221 0 0 0 1 0]\n",
" [ 0 0 13 0 2 102 0 6 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 1 0 0 0 0 108 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 1 1 0 0 0 0 0 65]]\n",
"\n",
"Epoch 112/119\n",
"----------\n",
"train Loss: 0.2313 Acc: 0.9222\n",
"Confusion Matrix:\n",
"[[239 1 0 9 3 0 8 1 0 2]\n",
" [ 3 260 0 2 9 0 0 0 1 6]\n",
" [ 0 0 180 1 0 16 0 9 0 0]\n",
" [ 14 1 0 195 3 0 3 0 0 0]\n",
" [ 0 13 0 1 369 1 3 0 2 3]\n",
" [ 0 0 6 0 0 187 0 12 0 1]\n",
" [ 4 2 0 4 0 0 161 0 0 0]\n",
" [ 0 0 3 0 1 5 0 173 0 0]\n",
" [ 0 3 0 2 6 0 0 0 148 0]\n",
" [ 0 1 0 4 1 0 0 0 1 115]]\n",
"val Loss: 0.1778 Acc: 0.9381\n",
"Confusion Matrix:\n",
"[[143 1 0 8 0 0 4 0 1 0]\n",
" [ 0 162 0 0 3 0 0 0 2 0]\n",
" [ 0 0 105 0 0 10 0 8 0 0]\n",
" [ 3 0 0 124 2 0 0 0 0 0]\n",
" [ 0 7 1 3 222 0 0 0 0 0]\n",
" [ 0 0 9 0 2 108 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 113/119\n",
"----------\n",
"train Loss: 0.2553 Acc: 0.9095\n",
"Confusion Matrix:\n",
"[[240 3 0 15 0 0 4 1 0 0]\n",
" [ 1 261 0 0 12 0 0 1 2 4]\n",
" [ 1 0 178 1 0 19 0 7 0 0]\n",
" [ 11 1 1 194 5 0 3 0 0 1]\n",
" [ 1 19 0 4 359 2 2 0 1 4]\n",
" [ 0 0 14 0 2 178 0 12 0 0]\n",
" [ 5 0 0 2 2 0 162 0 0 0]\n",
" [ 0 0 2 1 0 5 0 174 0 0]\n",
" [ 1 5 0 2 9 0 0 0 142 0]\n",
" [ 1 3 0 0 5 0 1 0 1 111]]\n",
"val Loss: 0.1852 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 163 0 0 1 0 0 0 3 0]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 3 0 0 123 2 0 1 0 0 0]\n",
" [ 0 7 1 3 220 0 0 0 1 1]\n",
" [ 0 0 16 0 2 100 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 5 0 1 0 0 1 0 0 65]]\n",
"\n",
"Epoch 114/119\n",
"----------\n",
"train Loss: 0.2504 Acc: 0.9177\n",
"Confusion Matrix:\n",
"[[243 1 0 14 0 0 3 0 0 2]\n",
" [ 1 264 0 0 12 0 1 0 0 3]\n",
" [ 1 0 178 0 1 15 0 10 1 0]\n",
" [ 12 2 0 190 7 0 2 1 1 1]\n",
" [ 2 9 1 4 368 2 2 0 2 2]\n",
" [ 0 0 7 1 1 190 0 7 0 0]\n",
" [ 4 0 1 2 2 0 161 1 0 0]\n",
" [ 0 0 7 0 1 8 0 166 0 0]\n",
" [ 2 4 0 1 4 0 0 0 148 0]\n",
" [ 0 5 1 1 5 0 0 0 1 109]]\n",
"val Loss: 0.1835 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[142 1 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 105 0 0 7 0 11 0 0]\n",
" [ 1 0 0 125 2 0 1 0 0 0]\n",
" [ 0 2 1 3 226 0 0 0 0 1]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 4 0 1 1 0 1 0 0 65]]\n",
"\n",
"Epoch 115/119\n",
"----------\n",
"train Loss: 0.2495 Acc: 0.9099\n",
"Confusion Matrix:\n",
"[[240 1 0 10 2 0 7 0 1 2]\n",
" [ 6 255 0 0 12 0 1 0 2 5]\n",
" [ 0 0 172 0 1 15 0 17 0 1]\n",
" [ 7 2 0 200 5 0 2 0 0 0]\n",
" [ 1 9 2 5 370 1 1 1 0 2]\n",
" [ 0 0 14 2 2 177 0 11 0 0]\n",
" [ 8 2 0 5 1 0 155 0 0 0]\n",
" [ 0 0 4 0 0 9 0 169 0 0]\n",
" [ 1 2 0 0 4 0 0 0 151 1]\n",
" [ 1 2 1 1 4 0 0 0 2 111]]\n",
"val Loss: 0.1835 Acc: 0.9312\n",
"Confusion Matrix:\n",
"[[144 1 0 5 0 0 7 0 0 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 106 0 0 11 0 6 0 0]\n",
" [ 3 0 0 121 2 0 3 0 0 0]\n",
" [ 0 3 1 4 224 0 0 0 0 1]\n",
" [ 0 0 14 0 2 104 0 3 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 3 0 104 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 1 3 0 1 1 0 2 0 0 64]]\n",
"\n",
"Epoch 116/119\n",
"----------\n",
"train Loss: 0.2607 Acc: 0.9126\n",
"Confusion Matrix:\n",
"[[233 1 0 20 1 0 6 0 1 1]\n",
" [ 1 263 1 1 9 1 0 0 1 4]\n",
" [ 1 0 173 1 1 24 0 6 0 0]\n",
" [ 14 4 0 189 4 0 3 0 1 1]\n",
" [ 1 11 3 3 366 3 1 0 2 2]\n",
" [ 0 1 10 0 0 188 0 7 0 0]\n",
" [ 3 1 0 7 0 0 160 0 0 0]\n",
" [ 0 0 4 1 1 7 0 169 0 0]\n",
" [ 0 5 1 1 3 0 0 1 147 1]\n",
" [ 1 0 1 0 1 0 1 0 0 118]]\n",
"val Loss: 0.1860 Acc: 0.9320\n",
"Confusion Matrix:\n",
"[[141 1 0 8 0 0 6 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 107 0 0 8 0 8 0 0]\n",
" [ 1 0 0 126 1 0 1 0 0 0]\n",
" [ 0 3 1 5 223 0 0 0 0 1]\n",
" [ 0 0 17 0 1 101 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 2 1 0 0 0 91 0]\n",
" [ 2 3 0 1 1 0 1 0 0 64]]\n",
"\n",
"Epoch 117/119\n",
"----------\n",
"train Loss: 0.2610 Acc: 0.9177\n",
"Confusion Matrix:\n",
"[[242 0 0 12 3 0 3 0 0 3]\n",
" [ 6 254 0 3 11 1 1 0 3 2]\n",
" [ 1 0 177 0 0 21 0 7 0 0]\n",
" [ 11 0 0 196 6 0 3 0 0 0]\n",
" [ 0 9 0 6 367 3 0 1 4 2]\n",
" [ 0 0 12 0 3 182 0 8 0 1]\n",
" [ 0 1 0 4 0 0 166 0 0 0]\n",
" [ 0 0 2 0 1 10 0 169 0 0]\n",
" [ 1 3 0 0 5 0 0 0 150 0]\n",
" [ 0 4 0 0 2 1 0 0 1 114]]\n",
"val Loss: 0.1837 Acc: 0.9366\n",
"Confusion Matrix:\n",
"[[141 1 0 7 0 0 7 0 1 0]\n",
" [ 0 159 0 0 4 0 0 0 3 1]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 2 1 3 226 0 0 0 0 1]\n",
" [ 0 0 13 0 2 104 0 4 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 0 0 0 0 93 0]\n",
" [ 0 4 0 1 0 0 2 0 0 65]]\n",
"\n",
"Epoch 118/119\n",
"----------\n",
"train Loss: 0.2835 Acc: 0.9031\n",
"Confusion Matrix:\n",
"[[237 0 0 12 4 1 8 0 0 1]\n",
" [ 5 257 0 2 10 0 1 0 4 2]\n",
" [ 1 0 176 0 0 21 0 8 0 0]\n",
" [ 10 0 1 192 7 3 3 0 0 0]\n",
" [ 4 10 0 3 369 2 1 0 2 1]\n",
" [ 0 0 25 1 1 164 0 14 0 1]\n",
" [ 4 0 1 8 0 0 158 0 0 0]\n",
" [ 0 0 4 0 0 5 0 173 0 0]\n",
" [ 1 5 0 0 5 0 2 0 146 0]\n",
" [ 1 3 0 0 5 0 0 0 0 113]]\n",
"val Loss: 0.1851 Acc: 0.9374\n",
"Confusion Matrix:\n",
"[[141 2 0 7 0 0 6 0 1 0]\n",
" [ 0 160 0 0 4 0 0 0 3 0]\n",
" [ 0 0 104 0 0 7 0 12 0 0]\n",
" [ 2 0 0 124 2 0 1 0 0 0]\n",
" [ 0 2 1 1 229 0 0 0 0 0]\n",
" [ 0 0 12 0 2 104 0 5 0 0]\n",
" [ 0 0 0 0 0 0 101 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 0 4 1 1 1 0 0 0 0 65]]\n",
"\n",
"Epoch 119/119\n",
"----------\n",
"train Loss: 0.2452 Acc: 0.9181\n",
"Confusion Matrix:\n",
"[[240 2 0 12 0 0 6 0 0 3]\n",
" [ 5 263 0 1 6 0 0 0 0 6]\n",
" [ 0 0 176 0 0 22 0 8 0 0]\n",
" [ 6 1 1 202 4 0 1 0 0 1]\n",
" [ 0 11 2 5 366 3 0 0 3 2]\n",
" [ 2 0 14 0 2 178 0 10 0 0]\n",
" [ 3 0 0 3 1 0 164 0 0 0]\n",
" [ 0 0 3 0 0 9 0 170 0 0]\n",
" [ 0 3 0 1 4 0 0 0 151 0]\n",
" [ 2 3 1 2 5 0 1 0 0 108]]\n",
"val Loss: 0.1810 Acc: 0.9351\n",
"Confusion Matrix:\n",
"[[143 1 0 7 0 0 5 0 1 0]\n",
" [ 0 158 0 0 4 0 0 0 3 2]\n",
" [ 0 0 106 0 0 7 0 10 0 0]\n",
" [ 2 0 0 125 2 0 0 0 0 0]\n",
" [ 0 3 1 3 224 0 0 0 0 2]\n",
" [ 0 0 15 0 2 102 0 4 0 0]\n",
" [ 0 0 0 1 0 0 100 0 0 0]\n",
" [ 0 0 2 0 0 0 0 107 0 0]\n",
" [ 0 1 0 1 1 0 0 0 92 0]\n",
" [ 1 3 0 1 0 0 0 0 0 67]]\n",
"\n",
"Training complete in 55m 38s\n",
"Best val Acc: 0.944232\n"
]
}
],
"source": [
"model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler,\n",
" num_epochs=120)"
]
},
{
"cell_type": "code",
"source": [
"seconds = time.time()\n",
"print(\"Time in seconds since beginning of run:\", seconds)\n",
"local_time = time.ctime(seconds)\n",
"print(local_time)"
],
"metadata": {
"id": "Gw__FuGYx0Rc",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "41410466-2b3f-467c-cfc2-d10547d18b21"
},
"execution_count": 20,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Time in seconds since beginning of run: 1685210224.6163301\n",
"Sat May 27 17:57:04 2023\n"
]
}
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"id": "BD6f3RK7W2lE",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 923
},
"outputId": "cc4ba2cf-f9a8-4161-aa78-1ca2ff81114c"
},
"outputs": [
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAO0AAACXCAYAAAD59NwoAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAA9GUlEQVR4nO1deZAc1X3+eq6ee3Z29l7tvVqQViKIQ+IIhyIICAcVAYExhQ0xGBLAmFQqKVzlChAn2I6J7QApKnIKTBSbhKswJFzBAoyMFMkSEkhCq5VWu6u9Z4+Znft8+WP1e/u6p2cvrcBr3lc10kzP6+43vf317/49hTHGICEhsWRg+rwnICEhMT9I0kpILDFI0kpILDFI0kpILDFI0kpILDFI0kpILDFI0kpILDFI0kpILDFI0kpILDEsOdK+9957UBQF7733Ht92++23o7Gx8XObkx5Gc1yqePjhh6Eoymk/T2NjI26//fbTfp7fByw50i4mHn30Ubzyyiuf9zQKcPvtt0NRFHi9XiQSiYLvOzs7oSgKFEXBY4899jnMUKK7u5v/DWZ7dXd3AwD+4R/+AZs2bUJlZSUURcHDDz+8oHNbFu9nfH746U9/inw+P+/9Hn30UWzevBnXXXfd4k/qFGGxWBCPx/Haa6/hpptu0nz385//HHa7Hclk8rTP4zvf+Q4efPDB036ejo4OmExLR4aUl5dj69atmm3/9E//hL6+Pvz4xz8uGAtMXcuqqiqsWbMGb7311oLP/ZmRNp/PI51Ow263L/qxrVbroh/z84aqqrj44ovx3HPPFZD2F7/4Bb70pS/hpZdeOu3zsFgssFhO/22iquppP8diwuVy4dZbb9Vs+8///E9MTEwUbCccP34cjY2NGB0d5UReCOb1aCP75vDhw7jpppvg9XoRCATwrW99q+CprygK7rvvPvz85z9He3s7VFXFm2++CQDo7+/H17/+dVRWVkJVVbS3t+Ppp58uOF9fXx+uu+46uFwuVFRU4C//8i+RSqUKxhnZtPl8Hv/8z/+M1atXw263o7y8HFdffTV++9vf8vnFYjE8++yzXI0RbarFnmM8Hsfhw4cxOjo663Um3HLLLXjjjTcQCoX4tt27d6OzsxO33HKL4T6hUAgPPPAA6urqoKoqWltb8YMf/ECjiZBq99hjj2HLli1oaWmBqqo4//zzsXv3bs3xjGxa+tu+8sorWLVqFb8+9PcV8d577+G8886D3W5HS0sL/vVf/9XwmEY2bVdXF2688UaUlpbC6XTiggsuwP/8z/8UHF9RFDz//PN45JFHUFtbC4/Hg82bNyMcDiOVSuGBBx5ARUUF3G43/uzP/qzg7/PMM8/gj/7oj1BRUQFVVbFy5Uo89dRThtf3VLBYfpcFPUJvuukmNDY24nvf+x527tyJxx9/HBMTE/j3f/93zbht27bh+eefx3333YeysjI0NjZieHgYF1xwAf/Dl5eX44033sAdd9yByclJPPDAAwCARCKBDRs2oLe3F/fffz9qamqwdetWbNu2bU5zvOOOO/Czn/0MGzduxJ133olsNosPPvgAO3fuxHnnnYetW7fizjvvxNq1a3HXXXcBAFpaWgDgtMxx165dWL9+PR566KE52zLXX389/vzP/xwvv/wyvv71rwOYkrJnnnkmzjnnnILx8Xgcl112Gfr7+3H33Xejvr4eH374Ib797W9jcHAQP/nJTzTjf/GLXyASieDuu++Goij4x3/8R1x//fXo6uqaVXvZvn07Xn75Zdxzzz3weDx4/PHHccMNN6C3txeBQAAA8NFHH+Hqq69GdXU1HnnkEeRyOfzd3/3dnKTM8PAwLrroIsTjcdx///0IBAJ49tlnsWnTJrz44ov40z/9U834733ve3A4HHjwwQdx9OhRPPHEE7BarTCZTJiYmMDDDz+MnTt34mc/+xmamprwt3/7t3zfp556Cu3t7di0aRMsFgtee+013HPPPcjn87j33ntnnetnDjYPPPTQQwwA27Rpk2b7PffcwwCw/fv3820AmMlkYgcPHtSMveOOO1h1dTUbHR3VbL/55puZz+dj8XicMcbYT37yEwaAPf/883xMLBZjra2tDAB79913+fbbbruNNTQ08M/btm1jANj9999f8Bvy+Tx/73K52G233VYw5nTM8d1332UA2EMPPVRwPj1uu+025nK5GGOMbd68mW3YsIExxlgul2NVVVXskUceYcePH2cA2A9/+EO+33e/+13mcrnYkSNHNMd78MEHmdlsZr29vYwxxvcNBAJsfHycj/vlL3/JALDXXnuNb6O/uQgAzGazsaNHj/Jt+/fvZwDYE088wbdde+21zOl0sv7+fr6ts7OTWSyWgmM2NDRo/hYPPPAAA8A++OADvi0SibCmpibW2NjIcrkcY2z6uq5atYql02k+9itf+QpTFIVt3LhRc54LL7xQc68wxvjfU8RVV13FmpubC7bPhC996UsFxzZCMBic871ghAVZ/vqnzze/+U0AwOuvv67Zftlll2HlypXiAwIvvfQSrr32WjDGMDo6yl9XXXUVwuEw9u7dy49VXV2NzZs38/2dTieXijPhpZdegqIoeOihhwq+my18cbrmePnll4MxNm+P4S233IL33nsPQ0ND2LZtG4aGhoqqxi+88AIuueQS+P1+zbyvuOIK5HI5/PrXv9aM//KXvwy/388/X3LJJQCm1NLZcMUVV3DNBADOOusseL1evm8ul8M777yD6667DjU1NXxca2srNm7cOOvxX3/9daxduxZ/+Id/yLe53W7cdddd6O7uxqFDhzTjv/a1r2m0g3Xr1oExxjUUcfuJEyeQzWb5NofDwd+Hw2GMjo7isssuQ1dXF8Lh8Kxz/ayxIPV4+fLlms8tLS0wmUzctU1oamrSfA4GgwiFQtiyZQu2bNlieOyRkREAQE9PD1pbWwtIdsYZZ8w6v2PHjqGmpgalpaWzjtXjs5rjXHHNNdfA4/Hgv/7rv7Bv3z6cf/75aG1tLbjWwFQo6OOPPy6qftK8CfX19ZrPROCJiYlZ56Xfl/anfUdGRpBIJNDa2lowzmibHj09PVi3bl3B9hUrVvDvV61aVXQ+Pp8PAFBXV1ewPZ/PIxwOczX+N7/5DR566CHs2LED8XhcMz4cDvNj/a5gUdyCxaSX+AQDwJ0ht956K2677TbDfc4666zFmNKC8bs2R1VVcf311+PZZ59FV1fXjJI6n8/jyiuvxN/8zd8Yft/W1qb5bDabDcexOXQgOpV9TweKzWe2eR47dgwbNmzAmWeeiR/96Eeoq6uDzWbD66+/jh//+McLCiWebiyItJ2dnRopevToUeTz+Vm9Y+Xl5fB4PMjlcrjiiitmHNvQ0IADBw6AMaZ5KHR0dMw6v5aWFrz11lsYHx+fUdoaPWw+qznOB7fccguefvppmEwm3HzzzUXHtbS0IBqNzjrvzwIVFRWw2+04evRowXdG2/RoaGgwvI6HDx/m3y8GXnvtNaRSKbz66qsaaf3uu+8uyvFPBxZk0/7Lv/yL5vMTTzwBALPaKmazGTfccANeeuklHDhwoOD7YDDI319zzTUYGBjAiy++yLfF4/GiKquIG264AYwxPPLIIwXfiZLA5XJpwimnc44LCfkQ1q9fj+9+97t48sknUVVVVXTcTTfdhB07dhgG7kOhkMaOO90wm8244oor8Morr2BgYIBvP3r0KN54441Z97/mmmuwa9cu7Nixg2+LxWLYsmULGhsbNb6SU50noL0vwuEwnnnmmUU5/unAgiTt8ePHsWnTJlx99dXYsWMH/uM//gO33HIL/uAP/mDWfb///e/j3Xffxbp16/CNb3wDK1euxPj4OPbu3Yt33nkH4+PjAIBvfOMbePLJJ/G1r30Ne/bsQXV1NbZu3Qqn0znrOdavX4+vfvWrePzxx9HZ2Ymrr74a+XweH3zwAdavX4/77rsPAHDuuefinXfewY9+9CPU1NSgqakJ69atOy1zXEjIh2AymfCd73xn1nF//dd/jVdffRV/8id/gttvvx3nnnsuYrEYPvnkE7z44ovo7u5GWVnZvM59Knj44Yfx9ttv4+KLL8Zf/MVfIJfL4cknn8SqVauwb9++Gfd98MEH8dxzz2Hjxo24//77UVpaimeffRbHjx/HSy+9tGjZU3/8x38Mm82Ga6+9FnfffTei0Sh++tOfoqKiAoODg4tyDsLWrVvR09PD7eZf//rX+Pu//3sAwFe/+tW5aw/zcTWT+//QoUNs8+bNzOPxML/fz+677z6WSCQ0YwGwe++91/A4w8PD7N5772V1dXXMarWyqqoqtmHDBrZlyxbNuJ6eHrZp0ybmdDpZWVkZ+9a3vsXefPPNWUM+jDGWzWbZD3/4Q3bmmWcym83GysvL2caNG9mePXv4mMOHD7NLL72UORwOBkATcljsOS405FMMRiEfxqbCIt/+9rdZa2srs9lsrKysjF100UXsscce4yGRYvsyxgrmWCzkY/S31YdtGGPsV7/6FVuzZg2z2WyspaWF/du//Rv7q7/6K2a322fd99ixY2zz5s2spKSE2e12tnbtWvbf//3fmjF0XV944QXN9meeeYYBYLt379Zsp98TDAb5tldffZWdddZZzG63s8bGRvaDH/yAPf300wwAO378eMHvLIbZQj6XXXYZA2D4Eu+V2aAwNnfPwcMPP4xHHnkEwWDwM31iS/x+4brrrsPBgwfR2dn5eU9lSWLpZGhLLEnoq5Q6Ozvx+uuv4/LLL/98JvR7gN+LKh+J3100Nzfj9ttvR3NzM3p6evDUU0/BZrMVDUtJzA5JWonTiquvvhrPPfcchoaGoKoqLrzwQjz66KMFCToSc8e8bFoJCYnPH9KmlZBYYpCklZBYYpCklZBYYpCOKADqDC1wKKNYURQwxjCTA0BBYT4zfRZdB/ptDAzKyTPpz6BAMcyRXrArgo6l319R+Lbp31p4DponjTGaU7ECEsbY9D4nx9DI+f6a1GfQH+t3FZK0nwP0N7sCRfNeTxY9GYzIIhLFiFDTx5+d8DPN71RRjNAK5k/cLyokaecIThzMfHPpCbSQc4gc0UhhA7JRfyt6bzKZNOTm0lx4ryc//yxIW/G8hnOd2lEjsWf63TPrKCf31x9fwhCStLOAbp7T3a57PtKNyGmxWGC1WmG1WmGxWKCqKlRV5R0c8vk88vk8MpkMUqkU0uk0MpkMMpkMcrkccrmc9vzzVLmVqckUfZhoftcMh/6srvHvCyRp54i53s7F7E+9BC4qkRVjwopEdTqd8Hg8cLvd8Hg8cDgcsNvtcLlcKCkpgc1mQyqVgslkQj6fx+TkJFKpFBKJBBKJBGKxGH8lEgmkUimkUink83lBo5iFQoxN28dFhxhcNUFVKXZNpZSdGZK088CpqG/FHFSGB9fBbDZDVVWUlJTA7/ejtLQUpaWl8Pv9KCsrQ0lJCVwuF5xOJxwOB4LBIEZHR9Hc3Ay73Y5QKATGGDKZDOLxOCYmJhCJRDA+Po5wOIyJiQmMjo4iHA4jHo8jm81qnUbTk9ZIY3rwzGRDF0AYpr+ekqxzg8yIwszeY8CYT0YXjcaZTKZ5eXf13mPRQ2u1WuF2u1FZWYmGhgY0NDSgoqICXq+Xk6akpATxeByjo6MYGRnB6Ogo4vE4HA4HHA4HnE4nli1bhrq6Ovh8PuRyOcTjcYTDYaTTaYyMjKC3txf9/f0YHh7G8PAwYrGYVvIWcYLptQf9d8U8zIXXYH74InuPJWmxOKQVx8zVkzt1HGN11GQywel0IhAIoKKiAk1NTVi5ciXq6uqQz+cxPj6O/v5+nDhxAhMTEwiFQkilUlAUBRaLBSaTCdlslheLM8bg8/lQVVWFuro6LFu2jB+bMYbBwUEcOHAAnZ2d6O/vx8DAAEKhEDKZDN+/2LyLObb4WJ10ZlNvCscVvUqFkKT9gmMhpBXBhDF6wgJzCLHoPLUWiwUejweVlZWora1FfX09Wlpa4PP5MD4+js7OTvT09GB0dBSJRAImkwkOhwMejwcejweKomBwcBD5fB5NTU1wOByIx+NIp9MaKVxaWorGxkasWLECtbW1CIVCOHr0KEZGRjA0NISenh4MDg4iHo/P2uBMfDjR+4J49BxutbnejJK0X3DMh7RGnk4j0s6kNs4Eq9UKn8/H29+0t7ejrKwMo6OjOHDgAI4ePYp0Og2PxwNVVXnfp1wuB8YY7HY7stkshoaGkM/n0dbWhrKyMu5hTiaTiEajyGazCIfDmJychMViQWVlJZqbm9HW1oby8nL09/fj008/RUdHB44dO4ZwOKzxNhdcI4GgImkVReGEn+t1mMuoLzJppSNqgZjPk64YgfUS1mw2w+PxoKqqCs3NzTjnnHPg8/nw8ccfY/fu3YhEIigpKUF1dTWcTiey2SwikQii0SiSySTy+Tyy2SxUVYXP50MoFMLw8DByuRycTidsNhsURYHb7YbD4YDP50M8Hkc0GsXY2BgGBwfR1dWFyy67DC0tLQgEAvB4PLDZbDh69CjGx8f5Q4JUXPF3iTFj/e9njBVPf5KZFfOClLRYmKTVb9dD33hMlDj8WIIKSYStra1Fa2srVq5cCafTiV27duHQoUOwWq2orKyE3+9HLpdDJpNBJBJBOBxGJBJBLpeDoijweDwoLy+H2WxGf38/8vk8LBYLSkpK+IqFqqrC4XAgl8vBZDLBbDaDMYZwOIzh4WHk83mcffbZ2LBhAywWC/bt24dPPvkEHR0dGB4eLmrnGtm2PCVSTIsscErr0zpnh5S0EjNirkTlY3T2HTCzakg2bG1tLZYvX4729nbkcjm8/fbbGBgYgNfrRWlpKWw2G5LJJA/dDA4OIp1O83BOPp+H1WpFOp2G2WyG2WzmpPX5fHC73Uin0wCAdDrNHVekOrvdbrhcLgSDQezfvx+RSAQbN27EunXr4PP5YLVaub2saceqzDPRUSdZpdyYHyRpTyM0avHUhoK0P5Kw1dXVaGxsRHt7OxRFwYcffoiBgQG43W6+LEUqleKe4VQqhWg0CpPJxCUoZUCFw2EuPRVF4QSzWCxgjCGbzcJms8Fms8FkMnFPcyqVgtPpRE1NDex2O7q7u/Hyyy/juuuuw4oVK5DNZpHNZpFMJjE2NqbNqEJhccXcLhKkajxPSNLOgGLSg83wnWacoP4amXPk9S0vL0d9fT0n7K9+9Sv09fWhtLQULpeLk8XhcMBms3G1VlEUThy73Y5cLodUKoVMJsPJbbfbOVljsRhXg0niBgIBlJaWwuFw8HG5XA4OhwNerxd9fX144YUXcOWVV2LFihVIJpOIRCLIZDIIh8PTSRiKYnhd5kRenf0heTwzJGkXiIUQVwQRqqysDMuWLcOqVavg9/vx/vvvo6+vj6vEwJRn2GKxQFEUbk/SQtmxWAwWi4WvmyTmFWezWeTzeZjNZqTTab7wdzAYRCKRQD6fRyqVgqqqsNlsfMX3ZDKJdDoNVVURCAQQCoXwzjvv4JJLLkF9fT1WrVqFTCaDzs5ORCIRAMZEEz3JdNH4tVBOxnolO+cNSdoZoA/vzHp/CXnDxWKVwBRhrVYrSkpKUFtbi/b2dtTU1GD37t04ePAgnE4nSkpKuE1qs9m4ussY43nCFJMlG9Zms8HpdIIxhnQ6zWO4JB0ZY7Barcjn88jlcpy0kUgEdrudZ3LlcjmYzWbu2HI6nRgaGsKbb76JL3/5y1i7di1SqRSSySSOHz/O26TqH00LrnZa0F5fHEjSzgLF4D3dVMUygPTlcvqxFosFfr8ftbW1WLFiBerr63HkyBHs2bMHNpsNXq8XZrOZ25u5XA7JZJKrySTd7HY7LyRIp9NQFAV2ux02mw2MMR4CIpLTMfP5PEwmE2w2G/L5PEZGRqAoCpxOJ1RVhcvlgqqqyGQy3P71+/0YGBjAnj17cP3112PNmjUIh8NIpVLo7e3lGgBBk8p50pYvuE5TTzkO6ZCaGyRpTwVFvKZGcVnaRp7c6upqtLa2oq2tDeFwGDt37kQ2m0UgEICqqtwmJcJOTk4iHo9zSUhJFFT5wxjjhQUkJYnQdrude5OBqfCSXQhzUZVPIpGA1+uFw+Hg45PJJJLJJMxmM/x+P/bt24f6+npccMEFCIVCiEajiEajCAaDRbOmxBK+YmEf/XhJ3+KQpJ0BxZS7okrfLIauyWSC2+1GTU0N6uvr0dzcDEVRsHv3bkxMTKC0tJSTlTHGUw+pjI7K7ex2O6+bJbvV4XDA5XIhn88jnU7zDCmr1cpVZ0CbUkjvaf9kMolYLMbVcZvNBlVVkcvlkE6nUVJSglQqhW3btqGqqgqtra3o7+/HxMQEEokEV8HF4+urgIq11dFfRonikKQtglMpyC7mfFJVVeMp9vl82LNnDzo6OnjmkRimodpXyv01m808pkqEIkeU0+nk8VuezMAYTCYTL45Pp9PIZrOcyCSlSYW22WxIp9OIRCIaqS9qCaWlpejt7cWLL76I9evXo6mpCUNDQ0gmk+jt7eUrws2UIaW5wJKh84Yk7SLCKO+YYDabuUe4trYW1dXVOHLkCPbv388dSGSTWq1WxGIxRKNR3nHCarVCVVUe6jGbzZy8FG9NJBI8BET/K4rCCZ7P57kH2WKxcLWZ7FwqxRO7W7hcLthsNlitVp495XK50N3dje7ublxzzTWIRCK8oH5gYECTMTVrzJY7k+cR2/2CQ5J2npipNUoxCUtk9Pl8KCsrQ0NDA8bGxrBz506kUimUlpZy8gBTNibVs1LrGJKwREIiIuUTUxGAPkmfVFu73c5TFzOZDJfalGxBDwdyUGUyGcRiMWQyGdjtdjidTpjNZp7HnE6ncejQIaxZswZr1qxBJBLhHuvR0VGeVlm8wZyYeCLJOh9I0s4TC1GbKffX7/dj2bJlUFUV27dvx9DQEAKBgCZGSsn/2WwWFotF0wdKb2uSY4mISBDDNhTWyWQyvCie9meM8aofcm6RJ5rCRiR1yZ4WY8LBYBBvv/02br75ZqxZswahUIhrCKQmA2LBgDYjTIRIbumImhmyWfkCUdAeRcG0qndS2pH31uVyIRAIoLq6Gg0NDejt7cWRI0fgcrk4AUiKkXSjfYmYJAHFRAh9aIlITuozqdSk2op1t2Q/k+pNpX4UvyU7mZxd0WgUoVAI2WyWawWqqqKzsxP/+7//C5vNhra2NlRUVMDv9/N5A4IHXbx+gt0tMT9ISVsEs2U8FUgDQW8W1WSKyZaXl2P58uUwm83Yv38/FEWB3+/nNio5gCh0Q6Sj8E0mk8Hk5CQni8/n48Qzm81wu93IZDI8+YKOa7VaefxVJDo5qaiIIJlM8iwpkuoAuJ0bi8UwOTmJfD4Pt9uNXC7H47ydnZ1Ys2YNli1bhsrKSgSDQS51NZeoGEHZjB8ldJCSdgYwGEhUAfoWM3p7lkhRWlqK+vp6+Hw+7N+/H4ODgwWkSyaTiMfjGscRxUspSYLU4HQ6zW1WOi/ZnaRmk2OJVFwiH70oK4psV7EDBtXeWq1W2O127rCiMFQymeTE9nq9SKVS2LNnD6xWK+rq6hAIBPgDieaova6SlqcCKWnnCaO0RKOeSKSuUheKtrY2TExM4NChQ5yQJNWIXGQziipwLBbjqjLZtOSUonguJWFkMhluD5OKnM/nucQj9VnchzzODodDo0bTQ4Eqiux2Oy/lSyaT3D6mPOdjx45hfHwcra2tOHHiBMbHxzE+Pl6wEjwAw1UUJOYOKWlngAKNqQoAhXbYDDq0w+FAIBBAc3MzysvLMTAwgEgkAo/Hw9VXRVGQSqX4eHpRiiKlHFLyhMvlgtvt5uSjEI3YwYKcUER8Iiepz2Lig6qqXLKSBKYEDgolqaoKt9vNPcjUAF1VVdjtdrjdbqRSKezfvx9erxft7e2ora2F3+8vGqcV7X6upchu5XOCJO2pghUSmVRen8+HZcuWoampCWNjYzh48CCA6S4WVH2TSCS4/UmOI7JrxUIAIrOoApN6S9lTdFyS1nQM0TFG/5OziWK4JPnpHCRxRZWdHExiPjNJ6P3796O3txeNjY3cvlVV9bP/m/yeQ5K2CPTSdcaxBgUDVDq3fPlyeL1e7NmzB4ODgxpnEFXYUCIDZUFRBhTZjF6vlyfxExHpnOl0GvF4nHt9RZIRiPi0nyjlyTalB4PdbudhIgoVkepNCRv0sBDnDgDDw8PYtWsXVFVFW1sbli1bBp/PN6eieKkuzx3Spp0FDLOXmDFoU/aoG0VlZSWWLVuG7u5uHDx4EIwxOJ1OAOCOJ4rHkppLDifKdnI4HNwGhW4u5FUm0ohjxGPabDa4XC5uP4ukFY9Fnmcx7ESguYkNzEkVF0NQnZ2dGB0dxfLly9HX14e+vj6EQiEkk8mCpnYSC4MkbRHMpXa2YB/BTqSewmazGXv37sXY2BjPQCIJRWo0kUMkE93g4ip4ImHoO1JzxQZq5IQi6UoPAJGootSk/SjRIpVKFbRLpYQNIj3NI51OczXb6XQiGAzio48+QnNzM5qamniP5lQqJWOyiwRJ2jnAMHWxSD4jrQxQU1ODxsZGDAwMoKenh9/Y5HQiu5PCObSNMqAoO8moNlcEtYmhxAw6jmjX8ikL8WOqzSV1mvKSyfNMY8RKILEIgUhMyRekXsfjcXz88ce4+OKLuW3r9/u1rWlmuH4Ss0PatHOA3oNcdNxJCed2u1FdXc0zhtLpNPx+P09GoEobygcm9ZIyoPS2q8bDimkJqygK9zaT04jK8UQpSxlM5EQim1RMUyTyihIbKCyxE5cZoWMB4E3hPB4PL5Z3Op1obGxERUUFz6HGLKaGxOyQpJ0H5qLcWSwWeL1elJWVIRqNYnh4mFfUUDxWJJXdbkdJSQl8Ph9PayTpqServn2N6IyiZuVivrLe8UTFAWSf0ovsYpKeZFdrGtMZPDzE7hj0UKBWr4cOHUIoFEJdXR0qKyvhcDim9j15nXiIh03FbRdztfnfd0jSFoFRjJZSF4tmSAn2bHl5OcbHxxGNRnkclFRPMd5JYRO32w2v18vzgo1WdNc7jvSqLhG0WE4vkVIkrbhkh5h5pT+HmEklZlSJRAem1PWKigqMjY2hq6sLpaWlqKyshNvtNlxNUCPZpck7J0jSLhBG95e40p3D4cDo6CjvakiEojisvp0MxT0BY9tVH6oRnVCihKbUR3E/+l+fzSV+JuKJDjBRraaXEfHodxCh/X4/stksurq6kM/nUV5ezpvQzZRsITE3SEfUIoCWe6QOizU1NVAUhdeVqqqKeDzOvbhkZ9KNrl/EWZSy5CQS7Uo9calSiGxl0X6l4+lzjwmibUrz0D8caAxlWomZU/rjULjKarWir68PsVgMNTU1CAQC3GkFTId82MlGxzIENHdI0hZBMedmgTwQyoFoxbvy8nKkUilMTk5yqUrxUqq2ETtEkKQjtZQ7bWDctoXICoA/LMQui6INS8eg//UOJNpO5zd6cPDfLlQH0Rjx+PRgiMfjMJlMCIVCCIVC8Pv9KCkpgdVqNcxF1h9fYmZI9XiRQL2MSUUNh8OIRqO8godanAIocPaQN5eykPQxVL1kBQq7ZIhSnFTvYt5goFBtFs+lXzxM3I9AqrI4B/pNYoHB+Pg4V+HJuQbA0PEkpe3cICXtLJjrSgKUuuj1eqGqKu9QaLVaeaM0cgKJrWPo5qf2LOTxJTKTmjtbOEYkkUhM+mzkXKLPADTSU3wwiOcTjyeq6UR4itcqytRympOTk4hEInA4HPD7/bwIwvAiS8wZkrQLgFEohBqeVVRUwG6380WY3W43AHBVmGKqtC+BSCfGbI1IQtJZnwVlNEcCSXKRfOLxxSUvKV5L40TCi8cWS/iIuHSuZDLJuz+Gw2HYbDZ4PB5ue08f6FT+Cl9cSNIuACKRiAhi2MZkMvGkf1oYiyQn1csS6Sh0AoDHb408vsB0kzaSbCJxxHCMPvWRYrEUuyWQk4mSK4Dp0I/e4aVPX6TfpFfTxXAQAITDYWSzWb6QNWkVEguHJO0CoVcxrVYr3G43b8UiNjbLZDK8RpUIRU4pADxeKrZ50Us7IheV8lESB8V89baqmGpIjdpEbzBpBwTKpCqmCoutW+k36R9c9BCiMVarla+MQJJW9JobXVPGmGzsNgskaRcBREKXywWv1wvGGI/PUucHWiMHAHfWkPOIFmsWPb9GIClGqw2IDwNRKtKcRClL+1IXRjEZg+xpceU9mocoURmb7l9F8wHAj0VjKOSjqiqi0ShXkUnSGnqIZZh2zpCknQGG95Gm0Fbbo8ntdvNG46lUipfVkS1LkjYejyOTySAajfLm5HSjA1rpKkpcIlg8HueVQpQ6KKYtKorCvdFU/ifa3nQ8YDqJgkI+JG1FFZf/dMF2FZ1VlNkFgKv91BQ9mUyip6eHpzKKnmlNiIdJ7/FcIUk7XxjcVyTtyFaNRCKamxaYbjZO2VGxWAzZbBaxWIw7oai8TW8j0mciXDqd5hlUtOQHtTslslG9biwW48cWG6KLEtZut/M1g8guFmO29AARvdOkFdDi1Hq7mmptyQlGbXJsNhtfugQwjs1K6s4MSdoimKu2RiojhTVKSkrQ39/PV2MnSScmTzidTphMJsRiMb6+DgBevifanYC2uJ2cQkSGyclJuFwuja1IhI1EIkgkElyai55hAJptHo8HZrOZr1QgqrF6x5g4H1KrrVYrLzukrCjKb6bURupASceTknVhkKQ9RdCNS2u4kh1HNzdJHLJrydYjclLDNUVRNI3FxYJ3MQtJJA3ZppS8wRhDIpHgzipqXZPNZuHxeAzVXSI7ZXORM0xM8hDDS2IMmVbbE21zMg0odJXNZhEKhQAAfr8fLpcLwWBw6vz6FfSkB2pOkKQ1wHx8IkRayidOp9MIhUI8mZ+kjrgAFqmk1FSNyCFW9+hVZPFF0lGMoYpOLH1PZJL0NE/xOGIMmDHGpSZ5tOlcVNRA8wHAbWhx2U1FmWq8TusHZTIZjI2NAQAvHNA/lGYtn5LQQJJ2IaA6UJ0zx2w2Y2JiAsPDw9xJRPakx+MBMN2IDQBXW8kBJdqMRqqpGKcVIbaUoe6IpHpTNwqS+GIWk6gCiyo5nZM81NlslqvxFN4ymUxcjaYezmIDuHQ6DZfLBZPJhPHxcSSTSZSUlKCkpITnYgNSRV4IJGkNMGsnFF1uoyjBTpw4gbGxMY23llRHan1KxGWMaXJy9aEe0UkjJuWLkpZUZNomeqBpNTxSzyn0Q1JZVHfFcxEByeFFYRxSeUljYIzxOtlUKsXX+aHcarL1I5EIQqEQvF4vJy0/n+TsvCELBmbAjBqb8AUVC5ANS428KemBblAqKCBi6StrRNKKxeyid1W/DwCN9KWHBDU3p4SGTCbD7Wc6l+hYEm1lqvv1er3cuUbOrlQqxSt13G437HY7fziQXUxqPknhXC6HwcFB3qVSXKhaYv6Qknae0Es/krIOhwOqqvKlM8gzC4Av1yE2I6fMITHRX7Qti8VoxVRFkoii3Sw+EBRF4bHaVCqFaDSKdDrNVxMgiCqyqDJTjDmZTGqavtntdr7YNDm9AMDlcnGzgGxtsnuPHTuGmpoa3ltZYuGQpJ0B+gW2jJLzRdKS1AGmPK7RaJSPpzgoAE2XRFHSiSQVwzN0XiKtOA6YtqnFJm6kggYCAaRSKYyMjCASiSAej/MiBv3vEWOxohOLygpJelMGF/0myi2mxBExe4qu2/DwMKLRKJe0EguHJO0coU8GENU7svcowQGY7hYhJlSQfUhE19fGigQSCSt+1qvGdC6xiRsdlzKwgCk7lfKAScqLmUwiROlfUlICp9PJCU37JZNJ3tScvOD69Eqy2anrYzab1RRLSCwMkrSnCCIgqZypVIovVkU3NX1P7VrIaSV2mqBj6ckrklZMrDCaA0la0Ra2Wq28ZxMVHKTTab7SgXheAAVtZIhgpEFQeiRJWFoak5xrFHOm8BcAfq7S0lJeYyxJu3BI0s4TenuW7FdqEUpOKLEmljHGwzsANKQVbVtR7TVSxQFowjai1BaJQIUCove6oqICmUyG27dkV+q9xyKZ9OmTYqoimQQOh4Ov60NFEmQmkCou2rdkGoiFEQUmB6RTeSZI0s4A8cYxkgsk/chbK3ZapJua1n8lr7K+yJ2OM1PIRySLXsqSp5ckIhGTYqiUm2y1Wnlr03Q6zQvf6Tz6ZuXF5kS/j8YTYbPZLF9tj0JFNBdyvNEYIq1htY/ErJCknQFzbTMjOoLIdiWi0GcA3B7Ur3ynd0IRifTk0RMLmHJq+f1+WK1WZDIZTExMYHJyki9cLY6jnsyJRIJLS/E8wLRzSq9q62tvybOcSCTAGON2bSKR4DnHNJbe07o/ok1rRFxJ5Zkh47QzoJiKqh8jticVG7RR0gORlpYBAaZXoRNVU30iBUEkqVjLarFYEAgEUFpayjOU6PxiDa2YVOFwOLj3WN+RkSB6ocWX+GAiMppMJrhcLng8Hv77ac6i40q8NkbONIm5Q0raGTCbs0TMSCKPKXmKs9ksotEoAPDCc4qpigkSZA9TCEiUpjQHOgdJLNGzW1tby+1pkp5OpxPJZBIWi4Xn+ooOLXG1A1GC69vUiNdBfDCJv53SMImctAgXOc1oLD0ESJWWWDgkaU8RREaSrFQkwBjj6/ZQhQ2NERMlTCYTMpkM7+IoHlckLuUUO51OHiutq6uD3+/naq7Y9I0kvNvtLliAyyivWb+6gVFYScy6ogcEfUfnpnOITjhKOCkoFIDsdbwQSNKeIkTbjlqIkjeYSEwdG6hsDtA2A6fUQFJLAeN4MLWsIZvZ4/FwTzSp4vRZtB1F+5kgftYXCojf00NJ/I4eNlR6R2WAVCxA6jn9NpoXkb1YjyiJuUGSdg6YqfcxESadTsNsNsPr9fIbVFQ1KReZ7Fgx7EOfjby1BAotUbWQzWbT9BEWUwwpRqrPTBIhxmJpjvrfRf/rM7BoPGkTk5OT/KGhJy09yMTSQXHNIIn5Q5L2FCCmKMbjcSQSCXi9Xk03QlIPKdxDpCKVlcI4ovNHPL7eg+x2u3nyvhibJUcT2ZdihY4+U0qcu5FH2uiz3tlGiRRUgE/xYFEKk4SmUI/D4eDJHcUkraTy7JCknQFGN5DeBiPbkfJ6qSMFSRMqeiebkro6UPEAdWzUq8LFJJG+kTnd/GJjNzGHmOYsFgPoX8XsWaPcZIo5R6NR7gkWV0eg3yP2cs5kMnyFgVQqpSGtlLjzhyTtHEEqstFNRp0Vw+EwqqqqeG8oanxGJKXQC3mabTZb0d7FepVUJJh+G0lYkmqUA0xeWiN7Vi9JixFVBBGN2sVStwp932Zqfk4JHIqiwOfzQVVVvtKCxMIhSbtA6FVH6nyoKApcLhcGBga4rUqqIe0jSkqx4ZuovupbjdIYvcps1Duq2GLRpOKK68yKklbv1QWmi+7FODFJVZK85ICihxQV0Ivn93q9AMAzwzQPisX5k3xhIEm7QIg3HanI2WwWJSUlaGhoQEdHh2YZDlpRQFRV9ckKgDZ5QpS4Rml/4veiFKa6XrFrIzU5z+VyPGwkklVM8NAfUxxHRBVtUvqdYlsacr6RjUu5z9FolGsc/MF38jiSvHODJO08oPcii9KInFH5fB5VVVU8PkqSRYxjUv4uOaRE8lMcVyzbE8kqdpkQUwtpLBGKXiT1xCqcXC7HVXb9g0A8n1hwzxjjzjZxmU76PUYJJLRNVVX4fD4kk0neZN1IBZeFAnODJO08UayIIJfL8XhlRUUFfD4fIpEI7wdMFTCiRCN7UFGm62upvxKRUkxS4HMQpZROKiYSCV42J56HJC8w3eycOkHq1WQ6Hh2fCgPo2FQfS2qyGIum+VP1UiaTQXV1NUpKSjA6OsoLKIpBSt3ZIUm7SCA1NJfLoaysDNXV1QgGgzyJnsrhrFYrJzgRi1bco4wo0TNMtiLV4uoX6AKmF3MmiU92JTAdRxb7WJEH2OVyFRBW9ECT5BTXA9I7teicYnYXtdahYom6ujq43W4Eg0HpLV4ESNLOA0ZpD3QTUjJ8Pp+Hy+VCS0sLDh06xAvgk8kk7Ha7hggAeEIEgSSuSCbRiSTaxKLaSqElyi0m9Zi6IorSXJTc+u4Z4rmJoORkE6WkSG56CJD3mtblJS95XV0dj+3OlgklKT07JGnniJlKB0gtpZ5QdrsdDQ0N8Pl8CIVCsNvtPK4pJudTiITqUI1qbPW9kEV1GpiSsrT8B6AtWCeC6NfloUZwYpcKsq3FmK2oclONruicEm1fkvAAeOudsbEx1NTUoKysDLFYDGNjY7ws0AiSsHODLM1bJCSTSYyMjKCvrw9DQ0PweDw444wzeA0pY4z3aDKZphfsop7BRDy9pxYA70xBxBb7S5lMJu4NpnVySNoWK6+j5uniKnuAtpxO/E7s20w2rOhIE7szKorCw1v5fB5nnHEGPB4PwuEwRkdHEQ6HDW1aSdi5Q0raeUITEsG0BM5mswgGg+ju7kYgEMCyZcuwYsUK7Nu3D6lUCk6nExMTE5rQCNnAwDQxjfo/0Tn1bWVoP5vNBo/Hwx1bmUyGr4AnEkxc0YAksr4YgLYR6UwmE69AElccIBuaHgyUMEEF+IlEAoFAAKtXr4bb7cbg4CDC4TAnrbRtFw5J2lOAZqnak5J0aGgIyWQSbrcbHo8H7e3t+L//+z9eMKBPZCDJRZJXn8pIPYcVReHkoWQL0U4lqWez2biXl6QygUhL+dD6InV9VpSoohPp6aEgJv2TN5pUd3K0rVmzBvX19QCm7OLR0VFMTk5Kwp4iJGnnCH6bGVThsKkVkZFIJDA0NITe3l5UV1ejoaEBq1evxoEDB3irUSrXo9RFinOKLWjIGSTm+FLMU7Q7xaJ4sj31KY4iASnBIh6P8wWyfD5fQYqk6L2mnsdifjTNlb5PpVJ8rna7HclkEj6fD2effTZUVUV/fz96e3vR19fH29MYXluJOUGSdhGRyWQwPDyMEydOoKysDHa7HcuWLcPKlSuxa9cuvq6NuGYsqbz6mKzYoM1ms8HtdnPbmM5F3SnEFjZUeE8gIlosFm47p9NpTkIxlxnQqv+0/i2p2WJcVoz/UgyXwlOZTAarV69GQ0MDMpkM+vr60N3djeHhYUPVWMZm5wfpiFoACm6ukxvy+TzC4TD6+vowMDCAYDAIn8+HdevW8d7Ddrudh1CodE4f1snlcojFYojFYtwbTbasmIUlNgyn9EUAmpI8YDrNMBaL8YWzqMWp6PyiYnx9eiYVQ4RCIUxMTPCFp8k2pzit2+1GKpWCz+dDc3Mz7HY7YrEYRkdH0dfXJ1XjRYKUtIsEIkkqlcLg4CCCwSBfBqS1tRVr167F22+/zfNxI5EIdxCJsU8qbBftTOp44XK5eB9hWluHCEi2raqqPLmCHgg0hkIyZHfq+1hRXjAdhxxNlApJDwEiLNmuJPFJdb7ooovQ0tKCWCyGrq4unDhxAoODgwXpizM1F5AoDknaBaJYHjJjDBMTE+jt7UVzczNGRkZQX1+Piy++GF1dXejs7ITD4UA0GsXk5CQAcClKqjI1/KYuECQdSerqvcMkmUVvNGOM5xu7XC4oytTC0DRXyilOJpPctqayOfIM08NCTJoQbV5qNUPzCofDWLlyJc4++2y43W50dXWho6MDHR0dmJiYKOgPRddRYn6QpD0NSKVS6OnpweHDh1FSUgJFUVBTU4Mrr7ySL4TlcDgQi8UQCoX44l1i6iKt7aooCg8PGSVMiL2nKF2RFm0Ws6Io/9nhcHCpSCmN+tpaghhqouZ0dF7KbsrlcvB4PDzEc+mll6K0tBQ9PT04ePAgPv30U3R3d/M2shKnDknaU4C4+LS+pC0UCuHjjz/msVdVVbF69WpceeWV+OUvfwnGphanomwjypDK5XLcu+twOOD1enkIBwBXWakQgTzIdA5aLCuRSPAkDqvVikgkAqfTyR1aRDjR4yyqv4xNLRgtxo6J/KQSM8bg9Xq5HX7ppZeira0N4+PjOHLkCD766CN8+umniEQi0pZdREjSLjLo5szlchgeHsb+/ftht9sRCAQQCARw+eWXY2RkBO+99x48Hg/sdjsmJye5WkpkoP/FldNJ0lHnCFpiRPTqms1mxONxANM9jqljBADeX4qcTHQefaiJOms4nU5N5Q7Ngb43m82IRCJYsWIFzj//fKTTaXR0dGDv3r04fPgwwuFw0bRF6TVeGCRpFwn6rCIK2wwMDGDPnj18mZD29nZcddVVGB4eRkdHB7xeL9xuN0/IF3OMyX4lx49Y8uZ2uzXSS1ybhyQfFTHQqgKMTXd9BKZrcynzyeFw8BXx0uk0PB4PX2CLpCup2UTYcDiMtrY2XHvttfD7/dizZw/27t2LTz75BKFQaE69oGQd7fwgSbtIKChbUxQwTGUL9fb2coKoqor29nbceuuteO655/DJJ5/A6/XC6/XyFe2ocByYjsdSPS6pqyQZiZjkDBKzpajBnNhFggoDqAmbw+EoaA5H3mAKKxFZ6UWOs1gshoaGBtx4442or69HR0cHdu3ahf3793PHE3Ay+URi0SBJu4igm1OBAlBKIKaI293dDWCKGHa7Heeddx7uuusubNmyBZ988glKSkrg8Xg0sVRyLpEEJtWWanEB8NAPhYJIwou2bjKZ5Ko35SKTak02MDUeJ7WXCuTpXNSBw+l0Ip/PIxQKob6+Hl/5ylfQ1taGY8eO4YMPPsBvf/tbjIyMaFViA85KGi8ckrSLCKVI1JEI2NPTg+3bt/MMpzPPPBN33303tm7dip07d8LpdPL0RlpCkjy7lPVEnmKygaleliQyvRdL7GKxGM+IymQyfMEssWUNHUvMdKIcZWp8TtJ6cnISZ5xxBm688UYsX74cfX19+PDDD/Hhhx9iYGBAs1aPkVosCXtqkKRdZBh1dwCAPJty/HR3d+Pdd99FNpvF+Pg41q5di29+85uorq7GW2+9hfHxcXg8HgQCAQDgfZ2o46NRq1VSW8lbTHMg21Zs8BaPx7mDSvyOCEq5yaJzi44ZjUZhNptx0UUX4YYbbkBlZSU6Ozvx4YcfYtu2bejq6uIElzh9kKRdRMx4s7Ip9TmVSuHYsWM8D1hRFFx66aW488470d7ejjfeeAOHDh3CyMgIPB4PvF4v79FE3RTFAgOxmJ3sU6q2IXtW32+KyG+z2XjWVDqd5qq5qqqaXOdwOIxkMon6+nps3LgRF110EfL5PA4cOIDt27dj+/bt6OzsnI7F6p9ZksOLCknaRYAmO0rjiypcEY5U5RMnTnAJms1mcc4552DDhg1Ys2YN3n77bWzfvh19fX3o7+/nRe5OpxPRaJTbmUQ8KuuzWq2IRqM8W4psVUq2AMDL66LRKD83AP4goGU3GWN8Wcrq6mpceumlOP/881FVVYVQKIRdu3Zhx44d2LVrF06cOFE0eUKBIh1RiwyFSV0GqtCj6VSgnPxHb9sysIJtlGlUWlqKs88+G5dccgnWr1+PtrY25HI5hMNh9Pf3Y9u2bdi7dy9GRkY48axWK7xeL6xWK0+AIGk8Pj7O84ctFgvcbjdXoynDirzRtNIddbUg9RgAvF4v2tracP755+Pcc89FQ0MDwuEwdu/ejY6ODuzcuRP79u1DMBjUdHQUf99Mt9ap3nSpkymZX0RI0mKRSQuDNV7FQKSO1CaTCR6PB21tbbjkkktw4YUXor29HRUVFbwDxMjICHbv3o3f/OY3GBkZ4b2WqFCAQjXxeByTk5Pwer3c2eR2uzEwMACr1YrS0lIA086seDyOaDTKVW6fz4empiacddZZOOecc7B8+XK+MsDo6Cg+/vhjvP/++9i+fTuOHDnCpbXmtwrXQJL29ECSFotDWjGVseCG1ZF26j/tWrCqqqKiogIrVqzA2rVrsXr1aqxatQpVVVVQVRXZbBaTk5NIJpOYmJjglTMDAwPo7e1FMBhEKBRCOBzmMVjq8jg8PIx8Pg+bzcbVaQDweDyora1Fc3Mzli9fjra2NtTU1PD+UYlEAmNjYxgbG8PBgwfx/vvvY8eOHejr69NU7FATgLliMW44SdovOBaLtPoPCgoT8TWxXHGXk8kQNpsNVVVVWLVqFdauXYs1a9agubkZZWVlfJ1X8cGQy+W4HRuJRDQF56FQiGdPURy4rKwMgUCAd/0vLS3lTicA/OEwNDSE7u5u3nFiz549OHDgACYnJzWN2QoIO4fcREnaU4MkLRaRtKITCloVuRhpRXtXTC90Op2oqKhAU1MTVq5cibPOOgvl5eU877iqqgqlpaWaNjV0PnIwUVkdOaP07VgB8AKFZDKJsbExDA8P49ixYzh27BhOnDiBgYEB9PX1YXR0VBPOMVT9i6BY3eyp3HiStF9wLLqkhYFdqx/ICvOV9fsqisKdTlVVVaisrITP54Pf70djYyOWL1+O8vJyVFRU8JpZqv6hWKxYlkdZUKRiT05Ocps5GAyiv78fPT096O/vx8jICKLRqGZBLTH+Ox9I0i4uJGlxekgrOpwK7FuA37HFiMuH6zoiUoqh3+/nRKYFv1RV5VVBVPFDhKUyPlq/dnJyEuPj45icnEQoFML4+DhCoRB3cIlrAREWk7TSEbVwSNLiMyZtsZ11jqqpt9MSV3RyiS/KZSZ7mEr2xAoeKqmj4gJqR0PbyKmkX4BL42hC8d+zkLYxkrQLh0yuOE0olodcbKxRAgLZuwV2JKZXZRfzhCn1ULRbC2xqgZzUetVIghrnDE/N51Sl7RdeSpwiJGlPM2a7uQ0JYCB1pzZrPbVi10QCvRcbw4kOLpHI+iblc53zXL+fTfWXWBgkaU8T5iqJihJ2HvuIktJoLSC9JDWUorPkTU+/nbkxGwMAZT56hsR8IUl7GqC/kRXdd3O+oZkRSQrVXc33RYhpVH1kNFbf2G0eU12UMRKzQ5J2kTDTDWkojYpgLnYfZV/ROPFzsXPPNEYz3sDJtBBIgp4+SNL+jmEhEutUpJx0Di09SNJ+wSHJuvQg47QSEksMcgEuCYklBklaCYklBklaCYklBklaCYklBklaCYklBklaCYklBklaCYklBklaCYklBklaCYklhv8HobePiaPGtGQAAAAASUVORK5CYII=\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
}
],
"source": [
"visualize_model(model_ft)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "IMlakxzzW2lE"
},
"source": [
"## ConvNet as fixed feature extractor\n",
"\n",
"Here, we need to freeze all the network except the final layer. We need\n",
"to set ``requires_grad = False`` to freeze the parameters so that the\n",
"gradients are not computed in ``backward()``.\n",
"\n",
"You can read more about this in the documentation\n",
"[here](https://pytorch.org/docs/notes/autograd.html#excluding-subgraphs-from-backward)_.\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {
"id": "PcSyPZ_7W2lE"
},
"outputs": [],
"source": [
"model_conv = torchvision.models.resnet18(weights='IMAGENET1K_V1')\n",
"for param in model_conv.parameters():\n",
" param.requires_grad = False\n",
"\n",
"# Parameters of newly constructed modules have requires_grad=True by default\n",
"num_ftrs = model_conv.fc.in_features\n",
"model_conv.fc = nn.Linear(num_ftrs, 10)\n",
"\n",
"model_conv = model_conv.to(device)\n",
"\n",
"criterion = nn.CrossEntropyLoss()\n",
"\n",
"# Observe that only parameters of final layer are being optimized as\n",
"# opposed to before.\n",
"optimizer_conv = optim.SGD(model_conv.fc.parameters(), lr=0.001, momentum=0.93)\n",
"\n",
"# Decay LR by a factor of 0.1 every 7 epochs\n",
"exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "ufSETcSxW2lE"
},
"source": [
"### Train and evaluate\n",
"\n",
"On CPU this will take about half the time compared to previous scenario.\n",
"This is expected as gradients don't need to be computed for most of the\n",
"network. However, forward does need to be computed.\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"id": "v7mm_hpfW2lF",
"outputId": "cc9be7fe-8102-459c-b086-f54c6b93f27b"
},
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Epoch 0/119\n",
"----------\n",
"train Loss: 1.9127 Acc: 0.3035\n",
"Confusion Matrix:\n",
"[[ 72 23 20 27 69 1 20 9 15 7]\n",
" [ 51 68 18 10 98 6 4 10 7 9]\n",
" [ 17 12 60 6 29 37 6 26 4 9]\n",
" [ 57 7 5 35 64 6 20 6 12 4]\n",
" [ 44 61 16 16 209 13 10 8 6 9]\n",
" [ 13 7 48 8 29 57 5 29 7 3]\n",
" [ 31 9 13 12 44 6 43 3 5 5]\n",
" [ 6 10 28 4 18 34 4 72 5 1]\n",
" [ 27 31 8 5 55 4 2 2 25 0]\n",
" [ 4 4 13 6 63 1 2 1 2 26]]\n",
"val Loss: 1.3954 Acc: 0.5187\n",
"Confusion Matrix:\n",
"[[ 16 19 0 82 6 0 23 3 6 2]\n",
" [ 0 109 0 15 30 0 7 4 2 0]\n",
" [ 0 9 45 4 9 7 2 40 1 6]\n",
" [ 0 10 0 105 7 0 6 1 0 0]\n",
" [ 0 44 1 27 147 0 6 3 2 3]\n",
" [ 0 12 23 13 4 13 1 57 0 0]\n",
" [ 0 9 0 12 3 0 77 0 0 0]\n",
" [ 0 1 11 0 0 1 2 93 0 1]\n",
" [ 0 31 0 3 6 0 0 0 55 0]\n",
" [ 0 14 0 7 29 0 1 2 0 19]]\n",
"\n",
"Epoch 1/119\n",
"----------\n",
"train Loss: 1.3803 Acc: 0.5055\n",
"Confusion Matrix:\n",
"[[127 24 3 38 38 1 19 4 6 3]\n",
" [ 19 137 3 8 81 3 4 4 5 17]\n",
" [ 6 3 82 1 15 51 2 39 0 7]\n",
" [ 54 13 5 77 29 4 29 1 2 2]\n",
" [ 21 65 8 19 235 4 13 1 8 18]\n",
" [ 5 5 47 2 16 90 3 32 1 5]\n",
" [ 32 2 2 18 18 2 95 1 0 1]\n",
" [ 1 3 27 0 2 19 1 128 0 1]\n",
" [ 5 28 2 2 32 0 1 1 88 0]\n",
" [ 3 13 3 1 43 2 3 2 0 52]]\n",
"val Loss: 1.1307 Acc: 0.6333\n",
"Confusion Matrix:\n",
"[[ 48 10 0 55 5 2 27 4 2 4]\n",
" [ 1 115 0 12 18 1 8 5 3 4]\n",
" [ 2 1 26 0 1 61 1 20 0 11]\n",
" [ 1 9 0 98 6 2 10 2 0 1]\n",
" [ 1 29 0 21 142 4 7 2 8 19]\n",
" [ 1 1 1 1 1 94 0 20 0 4]\n",
" [ 2 8 0 6 1 0 84 0 0 0]\n",
" [ 0 0 2 0 0 15 1 90 0 1]\n",
" [ 0 9 0 3 1 0 0 0 82 0]\n",
" [ 0 8 0 4 8 1 0 1 0 50]]\n",
"\n",
"Epoch 2/119\n",
"----------\n",
"train Loss: 1.2492 Acc: 0.5423\n",
"Confusion Matrix:\n",
"[[121 16 4 44 36 0 31 3 5 3]\n",
" [ 18 134 3 11 79 8 9 3 8 8]\n",
" [ 1 2 99 2 6 56 5 26 0 9]\n",
" [ 52 3 3 86 39 3 23 2 2 3]\n",
" [ 20 46 5 25 254 5 9 2 5 21]\n",
" [ 3 3 52 2 7 99 3 36 0 1]\n",
" [ 25 9 3 9 12 0 110 1 0 2]\n",
" [ 1 0 20 0 3 17 4 137 0 0]\n",
" [ 7 21 1 2 30 0 1 1 94 2]\n",
" [ 1 8 1 7 39 2 2 4 0 58]]\n",
"val Loss: 0.9906 Acc: 0.6684\n",
"Confusion Matrix:\n",
"[[ 88 3 4 32 3 0 23 1 1 2]\n",
" [ 6 98 2 8 35 1 11 2 2 2]\n",
" [ 2 0 97 0 0 14 0 8 0 2]\n",
" [ 14 3 3 92 8 1 6 1 0 1]\n",
" [ 5 12 3 17 177 3 4 0 2 10]\n",
" [ 0 0 56 1 1 61 0 3 0 1]\n",
" [ 6 4 2 5 3 0 81 0 0 0]\n",
" [ 0 0 40 0 0 13 1 55 0 0]\n",
" [ 0 3 0 2 8 0 0 0 82 0]\n",
" [ 3 1 4 3 16 0 1 0 0 44]]\n",
"\n",
"Epoch 3/119\n",
"----------\n",
"train Loss: 1.1635 Acc: 0.5778\n",
"Confusion Matrix:\n",
"[[156 17 3 42 13 1 26 0 3 2]\n",
" [ 16 153 2 9 73 2 5 3 10 8]\n",
" [ 3 2 100 0 13 55 0 28 0 5]\n",
" [ 41 4 2 102 41 0 17 2 4 3]\n",
" [ 16 48 10 31 241 7 5 1 14 19]\n",
" [ 0 4 56 1 12 106 0 26 0 1]\n",
" [ 29 4 2 14 16 0 105 0 0 1]\n",
" [ 1 0 18 1 3 35 1 122 0 1]\n",
" [ 5 12 0 1 20 1 0 2 116 2]\n",
" [ 2 2 1 1 37 3 3 2 2 69]]\n",
"val Loss: 0.9448 Acc: 0.6937\n",
"Confusion Matrix:\n",
"[[119 5 1 13 1 0 12 2 1 3]\n",
" [ 14 122 0 0 11 1 7 4 3 5]\n",
" [ 3 0 84 0 0 16 0 15 0 5]\n",
" [ 38 8 1 67 4 1 8 1 0 1]\n",
" [ 12 28 2 11 138 1 8 1 7 25]\n",
" [ 2 1 35 1 1 61 0 19 0 3]\n",
" [ 10 6 0 2 1 0 82 0 0 0]\n",
" [ 0 0 12 0 0 4 1 91 0 1]\n",
" [ 1 5 0 2 0 0 0 0 87 0]\n",
" [ 3 3 1 2 5 0 1 0 0 57]]\n",
"\n",
"Epoch 4/119\n",
"----------\n",
"train Loss: 1.1434 Acc: 0.5969\n",
"Confusion Matrix:\n",
"[[162 8 4 38 21 0 20 2 1 7]\n",
" [ 13 163 1 5 73 4 3 1 3 15]\n",
" [ 3 5 102 4 8 49 3 27 0 5]\n",
" [ 47 5 3 107 29 3 18 1 2 1]\n",
" [ 17 52 5 20 246 9 10 1 13 19]\n",
" [ 1 5 52 0 13 108 2 25 0 0]\n",
" [ 24 11 0 16 7 0 109 4 0 0]\n",
" [ 1 3 26 1 4 12 2 130 0 3]\n",
" [ 3 11 2 4 19 0 1 4 114 1]\n",
" [ 1 10 4 0 29 4 0 2 1 71]]\n",
"val Loss: 0.9364 Acc: 0.6746\n",
"Confusion Matrix:\n",
"[[120 13 1 5 2 0 11 1 3 1]\n",
" [ 6 136 0 0 11 0 5 3 5 1]\n",
" [ 2 3 69 0 1 34 0 11 1 2]\n",
" [ 52 11 1 44 11 0 7 1 2 0]\n",
" [ 9 42 0 6 156 1 2 1 12 4]\n",
" [ 2 3 27 0 2 82 0 6 0 1]\n",
" [ 16 7 0 1 3 0 74 0 0 0]\n",
" [ 0 1 17 0 1 14 0 76 0 0]\n",
" [ 1 5 0 1 1 0 0 0 87 0]\n",
" [ 6 13 1 1 11 0 1 0 0 39]]\n",
"\n",
"Epoch 5/119\n",
"----------\n",
"train Loss: 1.0794 Acc: 0.6119\n",
"Confusion Matrix:\n",
"[[150 12 1 37 24 0 28 2 7 2]\n",
" [ 17 164 1 7 62 3 3 3 9 12]\n",
" [ 3 4 106 0 14 47 2 25 0 5]\n",
" [ 36 3 4 122 24 3 20 2 2 0]\n",
" [ 14 54 2 23 254 7 11 1 7 19]\n",
" [ 3 6 38 1 9 117 5 26 0 1]\n",
" [ 22 5 0 12 11 1 118 0 0 2]\n",
" [ 0 1 23 0 2 17 0 136 0 3]\n",
" [ 3 13 1 2 15 0 3 2 120 0]\n",
" [ 5 4 5 2 37 4 1 4 2 58]]\n",
"val Loss: 0.9310 Acc: 0.6753\n",
"Confusion Matrix:\n",
"[[ 86 10 2 8 4 0 35 1 6 5]\n",
" [ 3 117 2 0 23 0 10 2 6 4]\n",
" [ 0 1 104 0 0 4 0 11 1 2]\n",
" [ 27 4 1 55 20 0 16 1 2 3]\n",
" [ 3 15 4 7 181 1 3 1 6 12]\n",
" [ 0 1 79 0 3 29 0 9 0 2]\n",
" [ 2 2 1 2 2 0 92 0 0 0]\n",
" [ 0 0 31 0 0 0 1 75 0 2]\n",
" [ 0 2 0 1 5 0 1 0 86 0]\n",
" [ 1 1 1 1 7 0 2 0 0 59]]\n",
"\n",
"Epoch 6/119\n",
"----------\n",
"train Loss: 1.0430 Acc: 0.6356\n",
"Confusion Matrix:\n",
"[[160 11 1 41 20 3 20 2 4 1]\n",
" [ 9 173 2 10 56 3 6 4 10 8]\n",
" [ 1 4 110 0 9 50 4 23 1 4]\n",
" [ 54 7 1 115 22 0 13 2 0 2]\n",
" [ 17 50 6 20 264 7 4 1 4 19]\n",
" [ 2 2 53 0 5 113 0 27 0 4]\n",
" [ 24 5 1 14 4 1 120 0 0 2]\n",
" [ 0 0 21 0 1 22 3 135 0 0]\n",
" [ 2 9 0 3 10 0 4 1 129 1]\n",
" [ 3 5 3 2 27 1 1 1 1 78]]\n",
"val Loss: 0.8635 Acc: 0.7173\n",
"Confusion Matrix:\n",
"[[ 94 5 1 27 3 1 18 3 3 2]\n",
" [ 4 97 0 1 43 4 7 4 5 2]\n",
" [ 1 1 46 0 1 55 0 17 0 2]\n",
" [ 5 2 0 100 15 1 3 1 0 2]\n",
" [ 3 9 0 12 193 4 2 1 1 8]\n",
" [ 0 0 8 0 1 104 0 8 0 2]\n",
" [ 7 2 0 9 4 0 79 0 0 0]\n",
" [ 0 0 1 0 1 23 0 84 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 1 1 2 9 1 1 0 0 56]]\n",
"\n",
"Epoch 7/119\n",
"----------\n",
"train Loss: 0.9526 Acc: 0.6588\n",
"Confusion Matrix:\n",
"[[165 8 2 31 27 0 23 3 3 1]\n",
" [ 14 162 4 2 83 1 3 0 7 5]\n",
" [ 3 1 106 0 10 60 0 21 0 5]\n",
" [ 32 6 1 126 24 1 22 0 0 4]\n",
" [ 7 43 6 10 283 6 8 1 16 12]\n",
" [ 3 2 37 0 6 136 1 19 0 2]\n",
" [ 29 2 1 13 6 0 118 0 0 2]\n",
" [ 0 0 14 0 2 21 0 144 0 1]\n",
" [ 1 11 2 4 16 0 0 1 123 1]\n",
" [ 1 5 0 2 23 1 2 2 1 85]]\n",
"val Loss: 0.8375 Acc: 0.7265\n",
"Confusion Matrix:\n",
"[[105 4 1 27 1 0 13 2 2 2]\n",
" [ 9 111 0 3 23 1 8 4 6 2]\n",
" [ 3 0 59 0 1 38 0 20 0 2]\n",
" [ 13 6 0 99 7 0 3 0 0 1]\n",
" [ 5 17 0 13 174 1 2 1 8 12]\n",
" [ 1 0 15 1 1 85 0 18 0 2]\n",
" [ 10 2 0 8 2 0 79 0 0 0]\n",
" [ 0 0 2 0 0 9 1 97 0 0]\n",
" [ 0 2 0 3 3 0 0 0 87 0]\n",
" [ 4 2 1 3 6 0 1 0 0 55]]\n",
"\n",
"Epoch 8/119\n",
"----------\n",
"train Loss: 0.9574 Acc: 0.6733\n",
"Confusion Matrix:\n",
"[[172 13 0 34 20 0 18 0 4 2]\n",
" [ 12 178 0 9 54 7 8 1 6 6]\n",
" [ 1 1 127 1 2 40 1 31 0 2]\n",
" [ 27 6 0 140 20 1 15 2 1 4]\n",
" [ 10 48 6 16 271 6 8 3 7 17]\n",
" [ 4 2 46 1 13 120 1 19 0 0]\n",
" [ 18 6 0 16 5 0 123 1 0 2]\n",
" [ 0 1 13 0 1 14 0 152 0 1]\n",
" [ 1 18 0 2 10 0 1 1 125 1]\n",
" [ 1 9 2 4 30 2 1 1 0 72]]\n",
"val Loss: 0.8191 Acc: 0.7265\n",
"Confusion Matrix:\n",
"[[112 3 2 17 3 0 14 2 2 2]\n",
" [ 7 111 0 2 27 1 7 4 6 2]\n",
" [ 2 0 76 0 0 26 0 17 0 2]\n",
" [ 25 4 0 86 9 0 4 0 0 1]\n",
" [ 6 13 1 10 183 2 3 1 4 10]\n",
" [ 0 0 36 0 1 70 0 14 0 2]\n",
" [ 12 1 0 7 3 0 78 0 0 0]\n",
" [ 0 0 6 0 1 8 0 94 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 4 1 1 2 9 0 1 0 0 54]]\n",
"\n",
"Epoch 9/119\n",
"----------\n",
"train Loss: 0.9646 Acc: 0.6697\n",
"Confusion Matrix:\n",
"[[174 13 1 24 18 1 25 0 2 5]\n",
" [ 11 180 1 5 56 4 5 8 2 9]\n",
" [ 2 2 126 0 8 35 2 27 0 4]\n",
" [ 37 4 3 118 39 0 10 1 2 2]\n",
" [ 12 40 5 16 283 4 7 2 6 17]\n",
" [ 1 2 43 2 8 126 1 22 0 1]\n",
" [ 20 5 3 12 5 0 124 1 0 1]\n",
" [ 1 1 23 0 1 26 0 128 0 2]\n",
" [ 3 8 0 0 14 1 0 0 132 1]\n",
" [ 2 4 1 3 24 1 1 3 2 81]]\n",
"val Loss: 0.8183 Acc: 0.7235\n",
"Confusion Matrix:\n",
"[[111 5 1 18 3 0 13 3 1 2]\n",
" [ 8 114 0 2 24 1 7 4 5 2]\n",
" [ 2 0 61 0 0 34 0 24 0 2]\n",
" [ 26 5 1 85 7 0 3 1 0 1]\n",
" [ 6 16 1 10 178 3 3 1 4 11]\n",
" [ 0 0 20 0 1 81 0 19 0 2]\n",
" [ 11 3 0 7 3 0 77 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 4 2 1 2 7 0 1 0 0 55]]\n",
"\n",
"Epoch 10/119\n",
"----------\n",
"train Loss: 0.9514 Acc: 0.6702\n",
"Confusion Matrix:\n",
"[[172 8 4 35 15 1 17 3 5 3]\n",
" [ 17 175 1 6 52 6 5 4 4 11]\n",
" [ 3 2 120 1 9 37 4 26 0 4]\n",
" [ 39 1 1 134 22 0 13 5 0 1]\n",
" [ 21 45 5 14 276 2 4 2 6 17]\n",
" [ 0 2 37 0 4 135 1 24 0 3]\n",
" [ 24 4 0 11 10 0 122 0 0 0]\n",
" [ 0 4 20 3 0 13 0 139 0 3]\n",
" [ 3 17 0 3 10 0 0 0 126 0]\n",
" [ 2 10 3 4 24 2 0 2 1 74]]\n",
"val Loss: 0.8185 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[104 3 1 22 1 0 20 3 1 2]\n",
" [ 9 112 0 2 24 1 9 4 4 2]\n",
" [ 2 0 66 0 0 30 0 23 0 2]\n",
" [ 16 5 1 95 6 0 5 0 0 1]\n",
" [ 7 17 1 10 178 2 4 1 3 10]\n",
" [ 0 0 22 2 1 77 0 20 0 1]\n",
" [ 6 0 0 3 2 0 90 0 0 0]\n",
" [ 0 0 3 0 0 8 1 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 1 3 7 0 2 0 0 53]]\n",
"\n",
"Epoch 11/119\n",
"----------\n",
"train Loss: 0.9447 Acc: 0.6833\n",
"Confusion Matrix:\n",
"[[182 8 1 24 16 1 21 3 5 2]\n",
" [ 11 195 3 8 40 7 3 4 5 5]\n",
" [ 3 3 121 0 7 46 0 20 0 6]\n",
" [ 38 7 0 127 25 0 11 3 3 2]\n",
" [ 14 36 5 18 280 7 8 2 5 17]\n",
" [ 0 3 29 0 13 130 1 26 2 2]\n",
" [ 25 3 0 11 5 0 124 1 1 1]\n",
" [ 0 2 17 0 0 16 1 144 0 2]\n",
" [ 5 14 1 4 12 0 0 0 122 1]\n",
" [ 2 8 0 6 22 1 1 2 3 77]]\n",
"val Loss: 0.8048 Acc: 0.7349\n",
"Confusion Matrix:\n",
"[[110 5 1 18 3 0 15 2 1 2]\n",
" [ 7 113 0 0 30 1 7 4 3 2]\n",
" [ 2 0 75 0 0 29 0 15 0 2]\n",
" [ 22 3 1 86 11 0 4 1 0 1]\n",
" [ 5 13 1 10 186 2 3 1 2 10]\n",
" [ 0 0 30 0 1 77 0 13 0 2]\n",
" [ 9 2 0 7 3 0 80 0 0 0]\n",
" [ 0 0 6 0 1 7 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 4 1 2 8 0 1 0 0 54]]\n",
"\n",
"Epoch 12/119\n",
"----------\n",
"train Loss: 0.9533 Acc: 0.6597\n",
"Confusion Matrix:\n",
"[[165 11 0 35 17 2 23 2 3 5]\n",
" [ 13 163 5 4 63 5 9 2 8 9]\n",
" [ 2 1 118 2 8 43 2 26 0 4]\n",
" [ 39 2 5 124 22 1 18 1 2 2]\n",
" [ 14 38 5 20 282 4 5 3 10 11]\n",
" [ 0 3 41 3 9 124 3 23 0 0]\n",
" [ 21 5 0 9 4 0 130 0 1 1]\n",
" [ 1 1 22 0 1 19 0 136 1 1]\n",
" [ 1 12 1 4 10 0 1 0 129 1]\n",
" [ 5 3 2 1 27 1 1 2 1 79]]\n",
"val Loss: 0.8066 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[113 5 1 21 1 0 12 2 0 2]\n",
" [ 9 113 0 2 25 1 7 4 4 2]\n",
" [ 2 0 76 0 0 31 0 12 0 2]\n",
" [ 21 4 0 94 8 0 2 0 0 0]\n",
" [ 6 13 1 11 184 3 2 1 2 10]\n",
" [ 0 0 31 1 1 79 0 9 0 2]\n",
" [ 13 1 0 7 3 0 77 0 0 0]\n",
" [ 0 0 11 0 1 13 0 84 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 3 1 3 7 0 1 0 0 53]]\n",
"\n",
"Epoch 13/119\n",
"----------\n",
"train Loss: 0.9949 Acc: 0.6574\n",
"Confusion Matrix:\n",
"[[175 8 0 33 18 0 19 2 6 2]\n",
" [ 11 172 0 8 58 7 6 5 7 7]\n",
" [ 1 2 121 2 9 44 2 22 0 3]\n",
" [ 39 8 3 127 16 1 18 2 1 1]\n",
" [ 10 48 5 16 273 4 6 3 11 16]\n",
" [ 0 0 42 1 8 125 1 25 0 4]\n",
" [ 25 6 0 11 6 1 118 0 1 3]\n",
" [ 0 1 8 1 3 24 1 143 0 1]\n",
" [ 3 18 0 3 15 0 0 0 119 1]\n",
" [ 3 12 2 2 27 0 2 2 0 72]]\n",
"val Loss: 0.8209 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[106 4 1 18 2 0 21 2 1 2]\n",
" [ 8 100 0 0 38 1 9 4 5 2]\n",
" [ 2 0 63 0 1 35 0 20 0 2]\n",
" [ 23 3 1 84 10 0 6 1 0 1]\n",
" [ 6 9 0 8 194 3 3 1 1 8]\n",
" [ 0 0 20 0 1 84 0 16 0 2]\n",
" [ 8 1 0 4 2 0 86 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 1 5 0 1 0 86 0]\n",
" [ 3 1 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 14/119\n",
"----------\n",
"train Loss: 0.9549 Acc: 0.6702\n",
"Confusion Matrix:\n",
"[[162 14 5 24 18 1 24 0 8 7]\n",
" [ 15 164 1 6 69 1 7 2 8 8]\n",
" [ 3 0 123 1 5 47 0 21 0 6]\n",
" [ 30 4 1 122 37 0 17 2 2 1]\n",
" [ 10 37 6 14 293 4 5 2 6 15]\n",
" [ 2 3 41 1 9 130 1 16 1 2]\n",
" [ 15 4 3 15 1 0 132 0 0 1]\n",
" [ 0 1 17 0 0 14 1 147 0 2]\n",
" [ 5 11 1 1 13 1 1 0 124 2]\n",
" [ 0 6 1 5 30 0 2 2 0 76]]\n",
"val Loss: 0.8278 Acc: 0.7341\n",
"Confusion Matrix:\n",
"[[107 4 1 19 4 0 14 3 3 2]\n",
" [ 5 106 0 0 35 2 7 4 5 3]\n",
" [ 1 0 65 0 0 33 0 22 0 2]\n",
" [ 16 2 1 86 18 0 3 1 0 2]\n",
" [ 4 8 1 8 192 3 3 1 2 11]\n",
" [ 0 0 19 0 1 86 0 15 0 2]\n",
" [ 8 2 1 8 5 0 77 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 0 1 2 8 0 0 0 0 59]]\n",
"\n",
"Epoch 15/119\n",
"----------\n",
"train Loss: 0.9260 Acc: 0.6765\n",
"Confusion Matrix:\n",
"[[170 12 2 30 17 0 22 2 4 4]\n",
" [ 12 175 0 7 59 3 6 5 5 9]\n",
" [ 0 0 122 2 2 43 4 30 0 3]\n",
" [ 41 2 2 123 28 0 16 1 1 2]\n",
" [ 10 36 3 22 282 6 7 0 10 16]\n",
" [ 0 2 30 0 8 136 1 29 0 0]\n",
" [ 17 4 1 12 7 1 129 0 0 0]\n",
" [ 0 1 16 1 0 17 0 146 0 1]\n",
" [ 3 15 1 2 10 0 1 0 126 1]\n",
" [ 0 7 1 4 25 2 1 1 3 78]]\n",
"val Loss: 0.8075 Acc: 0.7410\n",
"Confusion Matrix:\n",
"[[114 4 1 18 3 0 12 2 1 2]\n",
" [ 8 104 0 0 34 1 7 4 7 2]\n",
" [ 2 0 70 0 0 31 0 18 0 2]\n",
" [ 22 4 0 88 9 0 4 1 0 1]\n",
" [ 6 11 0 10 187 2 3 1 4 9]\n",
" [ 0 0 21 1 1 86 0 12 0 2]\n",
" [ 9 2 0 5 3 0 82 0 0 0]\n",
" [ 0 0 5 0 1 9 0 94 0 0]\n",
" [ 0 1 0 2 4 0 0 0 88 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 16/119\n",
"----------\n",
"train Loss: 0.9741 Acc: 0.6515\n",
"Confusion Matrix:\n",
"[[176 12 0 30 20 2 16 1 5 1]\n",
" [ 21 174 1 7 53 4 6 3 3 9]\n",
" [ 2 5 107 0 11 48 2 27 0 4]\n",
" [ 42 6 1 112 26 3 20 2 4 0]\n",
" [ 14 52 4 19 263 5 8 1 6 20]\n",
" [ 2 4 37 2 8 123 1 25 0 4]\n",
" [ 17 10 0 8 4 0 132 0 0 0]\n",
" [ 0 1 21 0 1 16 1 140 0 2]\n",
" [ 4 12 0 4 12 0 0 0 127 0]\n",
" [ 0 11 2 2 23 0 1 2 3 78]]\n",
"val Loss: 0.8100 Acc: 0.7296\n",
"Confusion Matrix:\n",
"[[108 7 1 19 2 0 12 1 5 2]\n",
" [ 8 112 0 0 31 0 7 3 4 2]\n",
" [ 2 0 65 0 1 36 0 16 1 2]\n",
" [ 22 3 0 90 10 0 3 0 0 1]\n",
" [ 6 13 0 8 190 1 3 1 1 10]\n",
" [ 1 1 21 1 1 83 0 12 0 3]\n",
" [ 10 3 0 7 3 0 78 0 0 0]\n",
" [ 0 0 7 0 1 13 0 87 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 1 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 17/119\n",
"----------\n",
"train Loss: 0.9552 Acc: 0.6765\n",
"Confusion Matrix:\n",
"[[168 15 2 30 24 1 15 0 5 3]\n",
" [ 15 186 0 3 55 2 6 4 4 6]\n",
" [ 2 2 124 0 5 45 5 22 0 1]\n",
" [ 36 8 2 127 21 1 15 1 4 1]\n",
" [ 11 41 3 18 284 2 7 3 8 15]\n",
" [ 0 5 44 3 7 117 0 25 2 3]\n",
" [ 13 4 2 10 5 0 135 1 0 1]\n",
" [ 0 2 18 0 0 23 1 137 0 1]\n",
" [ 5 7 1 3 6 1 0 0 135 1]\n",
" [ 2 13 1 3 25 0 1 1 2 74]]\n",
"val Loss: 0.8184 Acc: 0.7242\n",
"Confusion Matrix:\n",
"[[110 4 2 18 3 0 14 1 3 2]\n",
" [ 8 103 0 2 31 1 7 4 7 4]\n",
" [ 2 0 70 0 1 28 0 20 0 2]\n",
" [ 20 2 1 89 11 0 4 0 0 2]\n",
" [ 6 8 0 10 187 3 3 1 4 11]\n",
" [ 0 0 30 1 1 72 0 17 0 2]\n",
" [ 10 1 1 7 3 0 79 0 0 0]\n",
" [ 0 0 7 0 1 7 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 1 0 0 58]]\n",
"\n",
"Epoch 18/119\n",
"----------\n",
"train Loss: 0.9357 Acc: 0.6706\n",
"Confusion Matrix:\n",
"[[165 11 2 35 19 1 18 1 6 5]\n",
" [ 16 165 3 8 57 2 7 3 9 11]\n",
" [ 0 0 126 1 8 44 1 23 0 3]\n",
" [ 33 4 2 128 25 2 16 3 1 2]\n",
" [ 13 42 5 16 282 7 2 2 5 18]\n",
" [ 0 4 40 2 9 129 1 20 0 1]\n",
" [ 16 7 2 12 4 0 126 1 2 1]\n",
" [ 0 0 15 1 1 17 1 147 0 0]\n",
" [ 4 15 0 3 14 0 2 0 120 1]\n",
" [ 1 3 2 1 21 3 2 2 1 86]]\n",
"val Loss: 0.8105 Acc: 0.7349\n",
"Confusion Matrix:\n",
"[[111 3 1 21 1 0 15 2 1 2]\n",
" [ 9 110 0 2 26 1 8 4 5 2]\n",
" [ 2 0 67 0 0 29 0 23 0 2]\n",
" [ 15 4 1 95 9 0 4 0 0 1]\n",
" [ 6 13 1 11 180 2 3 1 6 10]\n",
" [ 0 0 23 1 1 77 0 19 0 2]\n",
" [ 10 0 0 7 2 0 82 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 1 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 19/119\n",
"----------\n",
"train Loss: 0.9579 Acc: 0.6656\n",
"Confusion Matrix:\n",
"[[175 15 2 27 18 1 16 1 3 5]\n",
" [ 12 174 2 9 60 1 7 6 5 5]\n",
" [ 1 2 126 3 5 38 3 23 0 5]\n",
" [ 43 5 2 122 21 0 17 0 3 3]\n",
" [ 11 48 0 18 272 8 7 2 5 21]\n",
" [ 0 3 36 0 8 128 2 24 1 4]\n",
" [ 18 2 0 17 5 1 126 0 0 2]\n",
" [ 0 1 17 0 0 18 3 142 0 1]\n",
" [ 7 14 0 5 8 0 1 3 119 2]\n",
" [ 2 4 1 4 26 0 3 2 1 79]]\n",
"val Loss: 0.8017 Acc: 0.7380\n",
"Confusion Matrix:\n",
"[[112 4 1 21 1 0 13 2 1 2]\n",
" [ 9 113 0 2 24 1 7 4 5 2]\n",
" [ 2 0 67 0 1 33 0 18 0 2]\n",
" [ 17 3 0 97 7 0 4 0 0 1]\n",
" [ 8 13 0 11 180 2 3 1 4 11]\n",
" [ 1 0 23 1 1 79 0 16 0 2]\n",
" [ 10 1 0 7 2 0 81 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 3 1 1 4 6 0 1 0 0 56]]\n",
"\n",
"Epoch 20/119\n",
"----------\n",
"train Loss: 0.9695 Acc: 0.6547\n",
"Confusion Matrix:\n",
"[[164 10 0 40 12 1 27 1 1 7]\n",
" [ 16 175 0 9 60 3 6 2 6 4]\n",
" [ 2 3 111 2 9 46 3 26 0 4]\n",
" [ 37 10 1 122 24 2 17 0 0 3]\n",
" [ 6 43 7 27 269 3 4 1 8 24]\n",
" [ 0 6 32 3 9 122 1 30 0 3]\n",
" [ 22 2 0 11 4 0 132 0 0 0]\n",
" [ 0 2 16 0 0 15 2 145 0 2]\n",
" [ 3 17 2 3 12 0 0 0 120 2]\n",
" [ 2 10 4 2 21 1 2 0 1 79]]\n",
"val Loss: 0.8086 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[114 4 1 18 2 0 13 2 1 2]\n",
" [ 11 106 0 1 26 1 8 4 7 3]\n",
" [ 2 0 69 0 0 32 0 18 0 2]\n",
" [ 22 5 1 88 8 0 4 0 0 1]\n",
" [ 6 15 1 10 177 3 4 1 5 11]\n",
" [ 0 0 22 1 1 83 0 14 0 2]\n",
" [ 10 1 0 7 2 0 81 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 3 1 1 3 6 0 1 0 0 57]]\n",
"\n",
"Epoch 21/119\n",
"----------\n",
"train Loss: 0.9610 Acc: 0.6520\n",
"Confusion Matrix:\n",
"[[161 14 2 34 18 1 25 0 4 4]\n",
" [ 13 171 2 11 60 1 6 3 6 8]\n",
" [ 3 2 114 4 6 50 3 24 0 0]\n",
" [ 34 8 2 122 22 1 20 3 3 1]\n",
" [ 15 47 3 16 271 9 6 2 4 19]\n",
" [ 3 2 34 2 9 134 1 20 0 1]\n",
" [ 16 6 1 15 3 0 128 0 0 2]\n",
" [ 1 2 23 0 1 20 1 132 0 2]\n",
" [ 4 15 1 5 10 0 0 0 124 0]\n",
" [ 1 8 0 3 30 0 0 3 1 76]]\n",
"val Loss: 0.8137 Acc: 0.7364\n",
"Confusion Matrix:\n",
"[[111 6 1 19 2 0 13 1 2 2]\n",
" [ 6 114 0 2 26 1 7 4 5 2]\n",
" [ 2 0 64 0 1 33 0 21 0 2]\n",
" [ 20 5 1 91 8 0 3 0 0 1]\n",
" [ 6 16 1 10 182 1 3 1 3 10]\n",
" [ 0 0 21 1 1 83 0 15 0 2]\n",
" [ 10 2 0 7 3 0 79 0 0 0]\n",
" [ 0 0 3 0 1 9 0 96 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 2 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 22/119\n",
"----------\n",
"train Loss: 0.9492 Acc: 0.6729\n",
"Confusion Matrix:\n",
"[[182 8 0 21 20 0 22 2 6 2]\n",
" [ 11 171 3 7 61 1 6 4 6 11]\n",
" [ 1 2 123 3 6 51 3 15 0 2]\n",
" [ 38 5 1 129 21 0 16 2 4 0]\n",
" [ 11 48 6 11 277 1 7 2 10 19]\n",
" [ 3 2 43 1 5 128 0 19 1 4]\n",
" [ 19 3 2 12 10 0 122 2 1 0]\n",
" [ 0 1 15 0 2 18 2 144 0 0]\n",
" [ 5 6 2 1 12 0 1 0 131 1]\n",
" [ 3 10 4 1 24 1 0 3 4 72]]\n",
"val Loss: 0.8100 Acc: 0.7296\n",
"Confusion Matrix:\n",
"[[107 7 1 21 1 0 15 1 2 2]\n",
" [ 5 116 0 1 26 1 7 4 5 2]\n",
" [ 1 1 67 0 1 31 0 20 0 2]\n",
" [ 20 6 0 90 7 0 5 0 0 1]\n",
" [ 5 21 0 10 179 1 3 1 4 9]\n",
" [ 1 1 25 1 1 76 0 16 0 2]\n",
" [ 8 3 0 7 2 0 81 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 1 4 0 1 0 87 0]\n",
" [ 3 2 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 23/119\n",
"----------\n",
"train Loss: 0.9576 Acc: 0.6688\n",
"Confusion Matrix:\n",
"[[165 6 3 38 21 2 18 2 7 1]\n",
" [ 18 167 3 9 59 3 7 2 5 8]\n",
" [ 5 0 117 1 5 37 4 32 1 4]\n",
" [ 41 4 4 124 24 2 15 1 0 1]\n",
" [ 9 49 7 12 279 4 5 1 7 19]\n",
" [ 1 5 30 3 6 139 1 21 0 0]\n",
" [ 16 4 0 7 7 2 134 0 1 0]\n",
" [ 0 0 16 0 3 13 1 148 0 1]\n",
" [ 2 19 0 5 7 0 1 2 122 1]\n",
" [ 3 8 2 3 29 0 1 1 0 75]]\n",
"val Loss: 0.8034 Acc: 0.7380\n",
"Confusion Matrix:\n",
"[[109 4 1 22 1 0 13 3 2 2]\n",
" [ 7 113 0 3 23 1 7 4 5 4]\n",
" [ 2 0 69 0 0 29 0 21 0 2]\n",
" [ 16 5 1 94 8 0 3 1 0 1]\n",
" [ 6 12 1 10 182 3 3 1 4 11]\n",
" [ 0 0 28 0 1 76 0 16 0 2]\n",
" [ 8 2 1 7 3 0 80 0 0 0]\n",
" [ 0 0 4 0 1 7 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 1 1 2 6 0 1 0 0 59]]\n",
"\n",
"Epoch 24/119\n",
"----------\n",
"train Loss: 0.9379 Acc: 0.6765\n",
"Confusion Matrix:\n",
"[[173 16 1 33 15 1 15 4 1 4]\n",
" [ 10 178 2 6 57 7 7 2 4 8]\n",
" [ 3 1 120 1 4 51 0 25 0 1]\n",
" [ 34 4 0 137 27 0 11 1 1 1]\n",
" [ 9 33 6 28 273 5 7 2 9 20]\n",
" [ 2 3 36 2 6 132 2 23 0 0]\n",
" [ 21 7 1 10 3 0 126 0 1 2]\n",
" [ 0 0 14 2 0 18 1 146 0 1]\n",
" [ 5 12 0 3 17 0 1 0 120 1]\n",
" [ 2 11 2 2 19 2 1 1 0 82]]\n",
"val Loss: 0.8058 Acc: 0.7395\n",
"Confusion Matrix:\n",
"[[110 4 1 23 1 0 12 2 2 2]\n",
" [ 7 111 0 3 28 1 7 4 4 2]\n",
" [ 2 0 67 0 0 37 0 15 0 2]\n",
" [ 15 4 0 96 9 0 4 0 0 1]\n",
" [ 6 12 0 10 185 2 3 1 5 9]\n",
" [ 0 0 22 1 1 87 0 10 0 2]\n",
" [ 10 1 0 7 3 0 80 0 0 0]\n",
" [ 0 0 6 0 1 10 0 92 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 1 3 7 0 1 0 0 54]]\n",
"\n",
"Epoch 25/119\n",
"----------\n",
"train Loss: 0.9498 Acc: 0.6702\n",
"Confusion Matrix:\n",
"[[171 11 1 30 16 2 18 1 7 6]\n",
" [ 14 183 2 5 48 3 6 4 7 9]\n",
" [ 0 1 119 2 12 51 1 17 0 3]\n",
" [ 33 5 2 133 22 2 15 2 0 2]\n",
" [ 14 37 2 18 286 6 6 1 6 16]\n",
" [ 2 5 44 1 4 120 0 27 0 3]\n",
" [ 19 7 1 11 5 0 125 2 0 1]\n",
" [ 0 0 27 0 0 21 4 130 0 0]\n",
" [ 0 10 0 5 15 0 0 2 127 0]\n",
" [ 2 5 1 0 30 0 3 2 0 79]]\n",
"val Loss: 0.8068 Acc: 0.7357\n",
"Confusion Matrix:\n",
"[[107 8 1 19 3 0 16 1 0 2]\n",
" [ 4 118 0 0 26 1 7 4 3 4]\n",
" [ 1 1 67 0 0 37 0 15 0 2]\n",
" [ 21 5 1 82 14 0 4 1 0 1]\n",
" [ 4 15 0 8 187 1 3 1 1 13]\n",
" [ 0 1 20 0 1 88 0 10 0 3]\n",
" [ 8 3 0 6 3 0 81 0 0 0]\n",
" [ 0 0 5 0 1 12 0 90 0 1]\n",
" [ 0 2 0 2 6 0 0 0 85 0]\n",
" [ 1 3 1 2 6 0 1 0 0 58]]\n",
"\n",
"Epoch 26/119\n",
"----------\n",
"train Loss: 0.9481 Acc: 0.6752\n",
"Confusion Matrix:\n",
"[[175 11 2 37 14 0 19 0 2 3]\n",
" [ 12 178 2 6 52 4 7 3 8 9]\n",
" [ 3 0 134 3 9 32 2 19 0 4]\n",
" [ 34 6 2 126 23 2 21 1 0 1]\n",
" [ 8 47 5 17 281 7 4 1 8 14]\n",
" [ 2 1 33 2 10 128 1 24 0 5]\n",
" [ 18 3 3 15 8 0 121 1 1 1]\n",
" [ 0 1 21 1 3 19 1 134 0 2]\n",
" [ 2 13 1 4 11 0 2 0 126 0]\n",
" [ 2 3 3 6 21 2 3 0 1 81]]\n",
"val Loss: 0.8043 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[107 7 1 18 3 0 14 1 4 2]\n",
" [ 6 116 0 1 25 1 7 4 5 2]\n",
" [ 2 0 73 0 0 27 0 19 0 2]\n",
" [ 23 6 1 83 10 0 4 1 0 1]\n",
" [ 5 15 0 9 184 2 3 1 4 10]\n",
" [ 0 0 26 1 1 75 0 18 0 2]\n",
" [ 8 3 0 7 4 0 79 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 27/119\n",
"----------\n",
"train Loss: 0.9549 Acc: 0.6711\n",
"Confusion Matrix:\n",
"[[176 7 3 31 18 0 19 1 3 5]\n",
" [ 12 180 2 4 61 2 3 1 6 10]\n",
" [ 3 2 111 2 7 55 3 17 1 5]\n",
" [ 41 8 1 127 16 1 18 1 1 2]\n",
" [ 13 37 6 21 268 6 4 2 12 23]\n",
" [ 1 3 33 2 5 137 1 23 0 1]\n",
" [ 16 4 2 13 7 0 128 1 0 0]\n",
" [ 0 2 26 1 0 9 2 140 0 2]\n",
" [ 2 11 0 3 11 0 2 0 129 1]\n",
" [ 0 5 4 6 26 0 1 1 0 79]]\n",
"val Loss: 0.8016 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[114 4 2 16 3 0 12 2 2 2]\n",
" [ 8 115 0 1 24 1 7 4 5 2]\n",
" [ 2 0 69 0 0 32 0 18 0 2]\n",
" [ 27 7 1 81 7 0 4 1 0 1]\n",
" [ 6 14 1 9 185 3 3 1 3 8]\n",
" [ 0 0 26 0 1 82 0 13 0 1]\n",
" [ 8 3 1 6 2 0 81 0 0 0]\n",
" [ 0 0 6 0 1 9 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 3 1 2 7 0 1 0 0 54]]\n",
"\n",
"Epoch 28/119\n",
"----------\n",
"train Loss: 0.9509 Acc: 0.6788\n",
"Confusion Matrix:\n",
"[[170 10 1 33 14 0 24 3 4 4]\n",
" [ 12 183 2 12 49 3 6 2 4 8]\n",
" [ 2 1 134 2 0 42 2 18 0 5]\n",
" [ 29 5 2 129 30 0 17 1 1 2]\n",
" [ 16 41 7 19 271 3 9 2 8 16]\n",
" [ 0 2 39 2 8 133 1 18 1 2]\n",
" [ 20 4 1 10 3 0 133 0 0 0]\n",
" [ 0 0 16 0 1 17 1 146 0 1]\n",
" [ 5 14 1 1 14 1 0 1 121 1]\n",
" [ 3 2 4 2 33 1 1 3 1 72]]\n",
"val Loss: 0.8214 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[107 8 2 19 4 0 12 1 2 2]\n",
" [ 4 122 0 0 25 1 6 4 3 2]\n",
" [ 1 1 65 0 0 34 0 20 0 2]\n",
" [ 20 4 1 82 18 0 2 1 0 1]\n",
" [ 5 17 1 8 183 3 3 1 3 9]\n",
" [ 0 0 20 0 1 87 0 13 0 2]\n",
" [ 12 5 0 7 5 0 72 0 0 0]\n",
" [ 0 0 3 0 1 10 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 29/119\n",
"----------\n",
"train Loss: 0.9442 Acc: 0.6765\n",
"Confusion Matrix:\n",
"[[161 6 4 37 24 1 20 0 6 4]\n",
" [ 14 177 5 10 54 1 3 2 4 11]\n",
" [ 2 2 135 2 3 41 1 17 0 3]\n",
" [ 33 5 0 136 23 1 14 0 1 3]\n",
" [ 10 47 4 14 282 6 4 0 7 18]\n",
" [ 1 5 34 2 9 130 0 23 0 2]\n",
" [ 16 8 0 14 9 1 123 0 0 0]\n",
" [ 1 0 16 0 1 17 1 144 0 2]\n",
" [ 7 15 0 4 12 0 0 0 121 0]\n",
" [ 2 11 1 3 24 0 1 1 1 78]]\n",
"val Loss: 0.8085 Acc: 0.7273\n",
"Confusion Matrix:\n",
"[[115 3 1 21 1 0 11 2 1 2]\n",
" [ 8 108 0 3 26 1 7 4 8 2]\n",
" [ 2 0 68 0 1 33 0 17 0 2]\n",
" [ 18 6 0 93 7 0 4 0 0 1]\n",
" [ 7 15 0 12 179 2 2 1 5 10]\n",
" [ 2 1 27 1 1 74 0 15 0 2]\n",
" [ 11 1 0 7 2 0 80 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 2 1 3 7 0 1 0 0 54]]\n",
"\n",
"Epoch 30/119\n",
"----------\n",
"train Loss: 0.9506 Acc: 0.6761\n",
"Confusion Matrix:\n",
"[[174 12 0 33 10 2 28 0 2 2]\n",
" [ 19 173 0 6 61 1 5 4 3 9]\n",
" [ 3 2 123 5 9 35 1 25 0 3]\n",
" [ 36 8 0 126 26 2 15 0 2 1]\n",
" [ 10 53 7 18 269 8 9 0 7 11]\n",
" [ 0 4 35 2 7 135 0 19 1 3]\n",
" [ 17 8 0 10 5 0 128 1 2 0]\n",
" [ 0 1 14 0 2 14 0 151 0 0]\n",
" [ 2 12 0 6 8 0 0 0 130 1]\n",
" [ 2 5 2 3 26 4 0 3 0 77]]\n",
"val Loss: 0.8122 Acc: 0.7357\n",
"Confusion Matrix:\n",
"[[110 7 1 19 3 0 11 1 3 2]\n",
" [ 5 113 0 0 30 1 6 3 5 4]\n",
" [ 1 1 66 0 1 39 0 13 0 2]\n",
" [ 17 4 0 87 15 0 4 1 0 1]\n",
" [ 4 10 0 8 190 3 3 1 2 12]\n",
" [ 0 0 17 0 1 95 0 8 0 2]\n",
" [ 10 5 0 7 5 0 74 0 0 0]\n",
" [ 0 0 5 0 1 18 0 84 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 2 1 2 8 0 0 0 0 58]]\n",
"\n",
"Epoch 31/119\n",
"----------\n",
"train Loss: 0.9432 Acc: 0.6683\n",
"Confusion Matrix:\n",
"[[167 12 2 27 14 0 29 1 6 5]\n",
" [ 6 181 1 8 59 4 4 6 6 6]\n",
" [ 1 3 119 0 13 49 2 17 0 2]\n",
" [ 43 6 0 124 18 1 19 1 1 3]\n",
" [ 15 39 7 13 276 3 7 2 12 18]\n",
" [ 0 3 39 1 6 134 2 19 0 2]\n",
" [ 21 7 2 11 2 0 127 1 0 0]\n",
" [ 0 0 20 1 1 16 0 143 0 1]\n",
" [ 7 14 1 2 8 0 0 0 126 1]\n",
" [ 2 9 1 3 30 2 1 2 0 72]]\n",
"val Loss: 0.8082 Acc: 0.7364\n",
"Confusion Matrix:\n",
"[[110 7 1 20 2 0 12 1 2 2]\n",
" [ 7 111 0 1 29 1 7 4 5 2]\n",
" [ 2 0 66 0 1 36 0 16 0 2]\n",
" [ 20 3 1 87 13 0 4 0 0 1]\n",
" [ 6 13 0 8 187 3 3 1 2 10]\n",
" [ 0 0 21 0 1 90 0 9 0 2]\n",
" [ 10 3 0 7 3 0 78 0 0 0]\n",
" [ 0 0 4 0 1 12 0 92 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 32/119\n",
"----------\n",
"train Loss: 0.9484 Acc: 0.6724\n",
"Confusion Matrix:\n",
"[[171 13 6 31 17 0 19 1 1 4]\n",
" [ 13 172 1 14 56 2 7 4 5 7]\n",
" [ 2 1 124 1 12 44 1 19 0 2]\n",
" [ 45 6 1 122 25 1 13 0 2 1]\n",
" [ 8 46 8 13 285 2 6 1 4 19]\n",
" [ 0 4 36 2 11 129 0 23 1 0]\n",
" [ 19 6 2 13 3 0 128 0 0 0]\n",
" [ 0 0 17 1 0 19 0 145 0 0]\n",
" [ 4 13 1 2 14 0 2 1 122 0]\n",
" [ 1 5 4 4 25 0 0 2 1 80]]\n",
"val Loss: 0.8126 Acc: 0.7257\n",
"Confusion Matrix:\n",
"[[112 6 1 20 2 0 9 1 4 2]\n",
" [ 5 117 0 0 25 1 7 4 6 2]\n",
" [ 1 1 70 0 1 30 0 17 1 2]\n",
" [ 23 5 1 88 9 0 2 0 0 1]\n",
" [ 5 18 0 10 179 1 2 1 7 10]\n",
" [ 0 1 31 1 1 72 0 15 0 2]\n",
" [ 12 4 0 6 3 0 76 0 0 0]\n",
" [ 0 0 6 0 1 10 0 92 0 0]\n",
" [ 0 2 0 3 3 0 0 0 87 0]\n",
" [ 2 2 1 2 8 0 0 0 0 57]]\n",
"\n",
"Epoch 33/119\n",
"----------\n",
"train Loss: 0.9416 Acc: 0.6729\n",
"Confusion Matrix:\n",
"[[167 13 2 28 21 1 22 2 4 3]\n",
" [ 9 179 0 7 63 3 7 2 6 5]\n",
" [ 2 4 122 1 5 53 1 17 0 1]\n",
" [ 38 5 1 133 21 2 12 3 1 0]\n",
" [ 10 46 6 18 282 5 5 2 3 15]\n",
" [ 2 2 33 1 6 138 1 20 0 3]\n",
" [ 21 3 0 9 5 0 133 0 0 0]\n",
" [ 0 1 22 0 2 17 0 138 1 1]\n",
" [ 4 15 1 3 17 0 1 0 116 2]\n",
" [ 0 7 3 2 30 5 1 2 1 71]]\n",
"val Loss: 0.7974 Acc: 0.7395\n",
"Confusion Matrix:\n",
"[[110 6 1 18 3 0 15 1 1 2]\n",
" [ 8 112 0 1 28 1 8 4 3 2]\n",
" [ 2 0 75 0 0 29 0 15 0 2]\n",
" [ 22 5 1 88 8 0 4 0 0 1]\n",
" [ 6 13 0 10 186 2 3 1 1 11]\n",
" [ 0 0 29 0 1 78 0 13 0 2]\n",
" [ 7 3 0 5 2 0 84 0 0 0]\n",
" [ 0 0 5 0 1 11 0 92 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 34/119\n",
"----------\n",
"train Loss: 0.9375 Acc: 0.6770\n",
"Confusion Matrix:\n",
"[[167 9 3 32 23 0 21 0 6 2]\n",
" [ 11 176 3 10 57 4 6 2 7 5]\n",
" [ 3 1 119 4 6 47 2 23 0 1]\n",
" [ 40 6 4 123 22 2 16 2 1 0]\n",
" [ 10 41 8 17 285 7 5 3 1 15]\n",
" [ 2 3 34 3 8 135 1 18 1 1]\n",
" [ 18 7 1 12 6 0 127 0 0 0]\n",
" [ 0 0 17 1 1 17 2 142 1 1]\n",
" [ 4 9 1 4 15 0 1 0 125 0]\n",
" [ 3 6 1 6 15 1 1 0 0 89]]\n",
"val Loss: 0.8050 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[110 6 1 18 3 1 12 3 1 2]\n",
" [ 6 117 1 0 25 2 7 4 3 2]\n",
" [ 0 1 67 0 0 35 0 18 0 2]\n",
" [ 25 5 1 81 11 1 3 1 0 1]\n",
" [ 4 18 2 8 181 3 3 1 1 12]\n",
" [ 0 0 23 0 1 87 0 11 0 1]\n",
" [ 8 2 1 7 3 0 80 0 0 0]\n",
" [ 0 0 4 0 1 11 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 3 3 2 7 0 0 0 0 56]]\n",
"\n",
"Epoch 35/119\n",
"----------\n",
"train Loss: 0.9495 Acc: 0.6670\n",
"Confusion Matrix:\n",
"[[155 12 2 37 20 1 27 0 3 6]\n",
" [ 10 186 1 8 47 5 6 2 6 10]\n",
" [ 1 1 117 2 5 45 2 29 0 4]\n",
" [ 37 4 1 131 21 2 17 1 1 1]\n",
" [ 21 48 2 17 273 3 7 1 8 12]\n",
" [ 1 3 36 1 9 122 0 32 1 1]\n",
" [ 14 4 0 11 6 0 134 0 1 1]\n",
" [ 0 1 20 0 1 14 0 145 0 1]\n",
" [ 3 12 0 3 13 0 1 1 124 2]\n",
" [ 1 8 1 2 26 2 0 3 0 79]]\n",
"val Loss: 0.8072 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[107 5 1 19 3 0 16 2 2 2]\n",
" [ 5 114 0 0 27 2 7 4 6 2]\n",
" [ 1 1 66 0 1 30 0 22 0 2]\n",
" [ 20 4 1 86 12 0 4 1 0 1]\n",
" [ 4 15 1 9 185 2 3 1 4 9]\n",
" [ 0 0 22 1 1 79 0 18 0 2]\n",
" [ 10 3 0 6 3 0 79 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 36/119\n",
"----------\n",
"train Loss: 0.9437 Acc: 0.6715\n",
"Confusion Matrix:\n",
"[[167 13 3 36 17 2 15 1 3 6]\n",
" [ 16 183 3 7 51 1 5 5 3 7]\n",
" [ 1 0 128 1 4 38 3 24 0 7]\n",
" [ 40 8 3 123 23 2 13 0 2 2]\n",
" [ 7 46 4 21 278 9 5 1 3 18]\n",
" [ 0 3 32 0 9 134 0 26 0 2]\n",
" [ 26 7 1 10 5 1 120 1 0 0]\n",
" [ 1 1 19 0 0 16 1 143 0 1]\n",
" [ 2 14 2 4 15 0 0 0 121 1]\n",
" [ 3 6 1 4 24 0 4 0 1 79]]\n",
"val Loss: 0.8052 Acc: 0.7418\n",
"Confusion Matrix:\n",
"[[107 4 1 19 3 0 16 3 2 2]\n",
" [ 7 114 0 2 27 1 7 4 3 2]\n",
" [ 2 0 63 0 0 36 0 20 0 2]\n",
" [ 19 3 1 88 12 0 4 1 0 1]\n",
" [ 6 12 0 8 189 3 3 1 1 10]\n",
" [ 0 0 17 0 1 92 0 11 0 2]\n",
" [ 9 2 0 6 3 0 81 0 0 0]\n",
" [ 0 0 4 0 1 10 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 3 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 37/119\n",
"----------\n",
"train Loss: 0.9399 Acc: 0.6829\n",
"Confusion Matrix:\n",
"[[163 11 2 35 16 0 28 1 3 4]\n",
" [ 11 187 2 11 49 3 6 0 5 7]\n",
" [ 2 0 131 1 2 42 6 15 0 7]\n",
" [ 35 6 1 130 27 1 12 1 1 2]\n",
" [ 11 49 2 25 274 6 6 1 4 14]\n",
" [ 0 6 34 1 12 131 0 20 0 2]\n",
" [ 19 5 0 10 3 0 132 1 0 1]\n",
" [ 1 1 18 1 1 12 2 141 1 4]\n",
" [ 1 8 1 3 12 1 1 1 131 0]\n",
" [ 1 4 1 2 30 2 0 1 0 81]]\n",
"val Loss: 0.8160 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[110 3 1 22 1 0 14 3 1 2]\n",
" [ 10 111 0 2 24 1 7 4 5 3]\n",
" [ 2 0 64 0 0 34 0 21 0 2]\n",
" [ 19 4 1 92 8 0 3 1 0 1]\n",
" [ 6 15 1 12 175 3 4 1 4 12]\n",
" [ 0 0 18 0 1 85 0 17 0 2]\n",
" [ 10 1 1 7 2 0 80 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 1 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 38/119\n",
"----------\n",
"train Loss: 0.9650 Acc: 0.6597\n",
"Confusion Matrix:\n",
"[[161 11 2 34 18 1 23 2 6 5]\n",
" [ 14 182 0 6 49 7 5 3 9 6]\n",
" [ 1 4 107 1 12 53 0 21 0 7]\n",
" [ 34 4 2 134 22 1 13 0 3 3]\n",
" [ 13 51 2 21 270 8 5 1 7 14]\n",
" [ 2 8 39 2 5 120 1 26 1 2]\n",
" [ 12 6 1 16 4 0 130 1 1 0]\n",
" [ 0 1 12 1 1 23 1 142 0 1]\n",
" [ 4 13 1 2 12 0 0 1 125 1]\n",
" [ 0 4 4 5 25 1 2 2 0 79]]\n",
"val Loss: 0.8032 Acc: 0.7288\n",
"Confusion Matrix:\n",
"[[113 5 1 17 2 0 15 2 0 2]\n",
" [ 9 110 0 1 30 1 7 4 3 2]\n",
" [ 2 0 65 0 0 35 0 19 0 2]\n",
" [ 28 4 1 81 9 0 4 1 0 1]\n",
" [ 6 13 1 10 185 2 3 1 1 11]\n",
" [ 0 0 21 0 1 84 0 15 0 2]\n",
" [ 10 2 0 5 2 0 82 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 3 1 2 8 0 1 0 0 53]]\n",
"\n",
"Epoch 39/119\n",
"----------\n",
"train Loss: 0.9241 Acc: 0.6833\n",
"Confusion Matrix:\n",
"[[176 4 0 34 21 0 20 3 3 2]\n",
" [ 10 184 2 7 55 3 4 2 8 6]\n",
" [ 3 3 118 1 7 45 2 25 0 2]\n",
" [ 45 6 4 127 14 0 13 1 3 3]\n",
" [ 14 41 3 17 279 5 8 2 7 16]\n",
" [ 1 2 36 0 12 129 1 22 0 3]\n",
" [ 11 3 2 11 5 0 139 0 0 0]\n",
" [ 0 1 17 0 2 17 1 144 0 0]\n",
" [ 4 10 1 7 12 0 1 1 120 3]\n",
" [ 0 7 1 2 20 1 2 2 1 86]]\n",
"val Loss: 0.8023 Acc: 0.7364\n",
"Confusion Matrix:\n",
"[[109 7 1 17 3 0 15 1 2 2]\n",
" [ 6 114 0 0 29 1 7 4 3 3]\n",
" [ 2 0 63 0 1 33 0 22 0 2]\n",
" [ 21 4 1 82 15 0 4 1 0 1]\n",
" [ 5 11 0 8 188 3 3 1 1 13]\n",
" [ 0 0 21 0 1 87 0 12 0 2]\n",
" [ 8 2 1 6 3 0 81 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 40/119\n",
"----------\n",
"train Loss: 0.9299 Acc: 0.6874\n",
"Confusion Matrix:\n",
"[[176 11 1 27 19 1 21 2 3 2]\n",
" [ 10 179 2 6 55 4 5 6 8 6]\n",
" [ 1 2 129 1 3 42 1 23 1 3]\n",
" [ 41 5 1 130 25 1 9 2 1 1]\n",
" [ 7 42 2 22 286 6 5 0 6 16]\n",
" [ 1 4 40 1 9 132 0 18 1 0]\n",
" [ 19 3 1 10 6 0 131 1 0 0]\n",
" [ 0 1 20 0 0 18 1 142 0 0]\n",
" [ 4 15 0 3 13 0 1 1 121 1]\n",
" [ 2 5 1 4 21 1 2 1 0 85]]\n",
"val Loss: 0.8132 Acc: 0.7464\n",
"Confusion Matrix:\n",
"[[109 4 1 20 3 0 13 3 2 2]\n",
" [ 6 112 0 3 27 1 7 4 5 2]\n",
" [ 2 0 65 0 0 35 0 19 0 2]\n",
" [ 14 3 0 97 11 0 3 0 0 1]\n",
" [ 6 11 0 10 187 3 3 1 2 10]\n",
" [ 0 0 18 0 1 90 0 12 0 2]\n",
" [ 9 3 0 7 3 0 79 0 0 0]\n",
" [ 0 0 3 0 1 10 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 41/119\n",
"----------\n",
"train Loss: 0.9112 Acc: 0.6929\n",
"Confusion Matrix:\n",
"[[169 11 1 35 18 0 16 1 6 6]\n",
" [ 9 196 1 8 42 5 7 3 5 5]\n",
" [ 2 0 122 1 6 51 2 17 0 5]\n",
" [ 31 11 1 124 25 1 19 0 0 4]\n",
" [ 4 37 4 18 292 5 8 1 3 20]\n",
" [ 0 4 34 1 5 135 2 24 0 1]\n",
" [ 22 2 1 14 6 0 126 0 0 0]\n",
" [ 0 0 20 0 1 16 0 144 0 1]\n",
" [ 4 10 1 2 13 0 2 0 126 1]\n",
" [ 0 7 1 2 21 0 1 1 0 89]]\n",
"val Loss: 0.8127 Acc: 0.7288\n",
"Confusion Matrix:\n",
"[[113 4 1 20 1 0 13 2 1 2]\n",
" [ 9 109 0 2 26 1 7 4 7 2]\n",
" [ 2 0 67 0 0 33 0 19 0 2]\n",
" [ 21 4 1 88 9 0 4 1 0 1]\n",
" [ 6 14 1 10 180 2 3 1 5 11]\n",
" [ 0 0 23 1 1 79 0 17 0 2]\n",
" [ 10 2 0 7 2 0 80 0 0 0]\n",
" [ 0 0 4 0 1 10 0 94 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 2 1 3 6 0 1 0 0 57]]\n",
"\n",
"Epoch 42/119\n",
"----------\n",
"train Loss: 0.9343 Acc: 0.6802\n",
"Confusion Matrix:\n",
"[[162 13 3 36 16 3 19 2 5 4]\n",
" [ 10 186 3 8 43 2 7 5 10 7]\n",
" [ 4 0 125 1 5 44 2 22 0 3]\n",
" [ 33 7 1 124 26 2 18 1 2 2]\n",
" [ 11 25 6 17 300 5 5 2 6 15]\n",
" [ 2 3 44 3 4 122 1 26 0 1]\n",
" [ 19 5 0 14 10 0 122 0 0 1]\n",
" [ 0 1 17 1 0 17 1 144 0 1]\n",
" [ 4 17 0 2 8 0 1 2 123 2]\n",
" [ 4 5 1 1 18 3 0 2 1 87]]\n",
"val Loss: 0.8040 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[113 7 1 17 3 0 12 1 1 2]\n",
" [ 6 115 0 1 29 0 7 2 4 3]\n",
" [ 1 1 77 0 1 29 0 12 0 2]\n",
" [ 23 6 0 87 8 0 4 0 0 1]\n",
" [ 6 13 0 8 189 1 3 1 3 9]\n",
" [ 1 1 30 1 1 80 0 7 0 2]\n",
" [ 9 3 0 7 3 0 79 0 0 0]\n",
" [ 0 1 12 0 1 16 0 78 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 1 2 7 0 0 0 0 56]]\n",
"\n",
"Epoch 43/119\n",
"----------\n",
"train Loss: 0.9577 Acc: 0.6674\n",
"Confusion Matrix:\n",
"[[171 8 2 32 16 0 24 0 5 5]\n",
" [ 11 191 0 3 53 5 8 3 4 3]\n",
" [ 2 1 117 4 5 52 1 22 0 2]\n",
" [ 46 5 4 123 18 1 15 2 1 1]\n",
" [ 13 43 7 18 282 3 3 1 7 15]\n",
" [ 3 4 35 0 10 125 2 27 0 0]\n",
" [ 22 3 0 13 13 1 116 2 1 0]\n",
" [ 0 0 15 0 1 23 3 139 0 1]\n",
" [ 4 11 0 3 18 0 1 0 121 1]\n",
" [ 3 3 3 5 22 0 0 4 0 82]]\n",
"val Loss: 0.8005 Acc: 0.7418\n",
"Confusion Matrix:\n",
"[[111 4 1 20 2 0 14 2 1 2]\n",
" [ 8 108 0 2 30 1 8 4 4 2]\n",
" [ 2 0 75 0 0 28 0 16 0 2]\n",
" [ 18 3 1 91 11 0 4 0 0 1]\n",
" [ 6 12 1 10 187 2 3 1 1 10]\n",
" [ 0 0 26 1 1 80 0 13 0 2]\n",
" [ 9 2 0 6 2 0 82 0 0 0]\n",
" [ 0 0 6 0 1 7 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 2 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 44/119\n",
"----------\n",
"train Loss: 0.9226 Acc: 0.6806\n",
"Confusion Matrix:\n",
"[[169 13 2 32 15 0 21 2 6 3]\n",
" [ 13 180 2 5 56 2 8 1 9 5]\n",
" [ 6 1 116 1 6 51 1 21 1 2]\n",
" [ 40 9 1 131 15 0 17 1 0 2]\n",
" [ 6 48 7 18 278 3 6 1 5 20]\n",
" [ 2 2 38 0 5 140 1 17 0 1]\n",
" [ 19 3 0 14 5 0 130 0 0 0]\n",
" [ 0 4 14 0 0 22 1 141 0 0]\n",
" [ 4 11 1 4 11 0 0 0 128 0]\n",
" [ 0 4 1 4 24 0 3 2 1 83]]\n",
"val Loss: 0.8070 Acc: 0.7227\n",
"Confusion Matrix:\n",
"[[116 4 1 15 1 0 17 1 0 2]\n",
" [ 12 112 0 2 23 0 7 4 3 4]\n",
" [ 2 0 79 0 0 26 0 14 0 2]\n",
" [ 28 4 0 83 8 0 5 0 0 1]\n",
" [ 9 14 1 10 174 2 4 1 1 17]\n",
" [ 2 0 38 1 1 66 0 13 0 2]\n",
" [ 10 0 0 6 2 0 83 0 0 0]\n",
" [ 0 0 9 0 0 7 1 91 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 1 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 45/119\n",
"----------\n",
"train Loss: 0.9508 Acc: 0.6788\n",
"Confusion Matrix:\n",
"[[174 12 0 33 19 0 18 0 5 2]\n",
" [ 14 185 3 6 47 3 4 4 7 8]\n",
" [ 2 3 122 3 7 41 1 22 0 5]\n",
" [ 38 6 2 131 15 2 18 2 1 1]\n",
" [ 13 49 10 21 267 1 6 2 9 14]\n",
" [ 0 2 36 2 7 134 1 23 0 1]\n",
" [ 12 3 1 18 9 1 126 0 0 1]\n",
" [ 0 0 13 0 1 17 3 146 2 0]\n",
" [ 3 18 1 4 11 0 0 0 122 0]\n",
" [ 2 6 2 3 20 0 1 0 3 85]]\n",
"val Loss: 0.8066 Acc: 0.7265\n",
"Confusion Matrix:\n",
"[[111 7 1 18 2 0 13 1 2 2]\n",
" [ 5 120 0 0 24 1 7 3 5 2]\n",
" [ 1 1 73 0 1 28 0 16 1 2]\n",
" [ 26 7 1 81 9 0 4 0 0 1]\n",
" [ 5 19 0 9 182 1 3 1 4 9]\n",
" [ 1 1 31 1 1 74 0 12 0 2]\n",
" [ 9 4 0 7 3 0 78 0 0 0]\n",
" [ 0 1 8 0 1 9 0 89 0 1]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 3 1 2 8 0 0 0 0 56]]\n",
"\n",
"Epoch 46/119\n",
"----------\n",
"train Loss: 0.9511 Acc: 0.6706\n",
"Confusion Matrix:\n",
"[[179 13 3 24 15 0 20 3 5 1]\n",
" [ 15 178 1 11 49 1 5 4 6 11]\n",
" [ 2 5 111 0 4 48 2 28 0 6]\n",
" [ 41 6 1 124 19 1 19 0 4 1]\n",
" [ 10 48 4 19 279 4 5 2 9 12]\n",
" [ 0 4 33 2 7 137 1 20 0 2]\n",
" [ 21 6 2 8 4 1 127 0 0 2]\n",
" [ 0 0 22 0 2 19 0 139 0 0]\n",
" [ 3 15 0 2 10 0 1 1 127 0]\n",
" [ 2 8 1 4 29 0 2 3 0 73]]\n",
"val Loss: 0.7984 Acc: 0.7364\n",
"Confusion Matrix:\n",
"[[113 4 1 18 2 0 14 2 1 2]\n",
" [ 8 115 0 1 25 1 7 4 4 2]\n",
" [ 2 0 74 0 0 30 0 15 0 2]\n",
" [ 21 7 1 90 5 0 4 0 0 1]\n",
" [ 6 16 1 10 181 2 3 1 2 11]\n",
" [ 0 0 33 0 1 76 0 12 0 1]\n",
" [ 9 2 0 5 2 0 83 0 0 0]\n",
" [ 0 0 4 0 1 10 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 3 1 3 8 0 1 0 0 52]]\n",
"\n",
"Epoch 47/119\n",
"----------\n",
"train Loss: 0.9410 Acc: 0.6793\n",
"Confusion Matrix:\n",
"[[182 10 2 35 10 1 19 0 4 0]\n",
" [ 12 182 2 6 58 2 6 5 2 6]\n",
" [ 1 2 120 2 3 47 5 22 0 4]\n",
" [ 38 4 1 123 27 3 17 1 1 1]\n",
" [ 7 40 5 24 287 2 4 3 4 16]\n",
" [ 1 2 41 2 12 119 2 23 0 4]\n",
" [ 15 5 0 8 5 0 137 1 0 0]\n",
" [ 0 1 19 1 2 25 2 131 0 1]\n",
" [ 2 12 0 4 9 0 1 0 129 2]\n",
" [ 2 5 3 3 22 1 1 2 0 83]]\n",
"val Loss: 0.7973 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[110 6 1 19 2 0 14 1 2 2]\n",
" [ 8 107 0 2 33 0 8 3 4 2]\n",
" [ 2 0 70 0 1 34 0 14 0 2]\n",
" [ 19 4 0 93 8 0 4 0 0 1]\n",
" [ 7 13 0 11 184 2 3 1 1 11]\n",
" [ 1 0 23 0 1 84 0 12 0 2]\n",
" [ 9 2 0 6 2 0 82 0 0 0]\n",
" [ 0 0 6 0 1 13 0 89 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 3 1 1 4 8 0 1 0 0 54]]\n",
"\n",
"Epoch 48/119\n",
"----------\n",
"train Loss: 0.9450 Acc: 0.6697\n",
"Confusion Matrix:\n",
"[[160 15 1 40 18 0 17 2 6 4]\n",
" [ 14 184 1 11 49 3 3 4 4 8]\n",
" [ 2 5 124 1 6 39 1 22 1 5]\n",
" [ 36 4 1 131 23 2 14 3 1 1]\n",
" [ 12 35 7 20 282 5 1 1 7 22]\n",
" [ 0 3 43 0 13 124 2 19 0 2]\n",
" [ 22 3 1 10 6 0 127 0 1 1]\n",
" [ 0 1 19 0 0 22 1 137 0 2]\n",
" [ 2 9 0 4 16 0 2 0 126 0]\n",
" [ 1 6 3 2 30 1 1 1 0 77]]\n",
"val Loss: 0.8072 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[108 4 1 19 3 0 17 2 1 2]\n",
" [ 9 110 0 1 27 1 8 4 5 2]\n",
" [ 2 0 72 0 1 22 0 24 0 2]\n",
" [ 18 3 0 93 10 0 4 0 0 1]\n",
" [ 6 14 0 10 185 2 3 1 2 10]\n",
" [ 1 0 27 1 1 70 0 21 0 2]\n",
" [ 8 2 0 6 2 0 83 0 0 0]\n",
" [ 0 0 5 0 1 7 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 49/119\n",
"----------\n",
"train Loss: 0.9417 Acc: 0.6720\n",
"Confusion Matrix:\n",
"[[169 12 3 31 24 2 16 0 6 0]\n",
" [ 8 186 2 11 43 4 8 5 4 10]\n",
" [ 2 3 105 0 8 56 2 24 1 5]\n",
" [ 39 3 2 129 24 0 14 1 1 3]\n",
" [ 13 45 4 15 275 4 7 2 8 19]\n",
" [ 0 2 40 2 7 131 2 21 0 1]\n",
" [ 17 4 2 8 7 1 132 0 0 0]\n",
" [ 0 1 18 0 0 20 2 141 0 0]\n",
" [ 6 8 0 2 15 0 1 0 126 1]\n",
" [ 2 6 1 3 25 0 1 1 0 83]]\n",
"val Loss: 0.7975 Acc: 0.7395\n",
"Confusion Matrix:\n",
"[[106 4 1 20 3 0 17 3 1 2]\n",
" [ 7 115 0 1 26 1 7 4 4 2]\n",
" [ 2 0 69 0 0 31 0 19 0 2]\n",
" [ 16 5 1 93 8 0 4 1 0 1]\n",
" [ 6 13 1 10 186 2 3 1 1 10]\n",
" [ 0 0 25 1 1 80 0 14 0 2]\n",
" [ 9 2 0 7 3 0 80 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 50/119\n",
"----------\n",
"train Loss: 0.9410 Acc: 0.6720\n",
"Confusion Matrix:\n",
"[[176 11 1 30 13 0 21 1 5 5]\n",
" [ 17 176 1 6 55 3 7 5 4 7]\n",
" [ 0 2 121 0 9 44 3 22 2 3]\n",
" [ 42 4 1 122 24 2 17 1 1 2]\n",
" [ 15 47 6 14 275 2 8 1 10 14]\n",
" [ 0 6 35 0 8 131 2 24 0 0]\n",
" [ 15 6 1 11 11 0 126 0 0 1]\n",
" [ 0 0 9 0 2 22 0 146 0 3]\n",
" [ 1 8 0 5 15 0 1 1 128 0]\n",
" [ 3 14 0 2 25 0 1 1 0 76]]\n",
"val Loss: 0.8078 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[108 7 1 18 3 1 15 2 0 2]\n",
" [ 5 118 0 0 27 1 7 4 3 2]\n",
" [ 0 1 64 0 0 37 0 19 0 2]\n",
" [ 26 6 1 78 11 0 5 1 0 1]\n",
" [ 5 16 1 8 185 3 3 1 2 9]\n",
" [ 0 0 18 0 1 89 0 13 0 2]\n",
" [ 8 4 0 6 3 0 80 0 0 0]\n",
" [ 0 0 3 0 1 9 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 3 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 51/119\n",
"----------\n",
"train Loss: 0.9554 Acc: 0.6588\n",
"Confusion Matrix:\n",
"[[170 11 3 33 17 1 17 0 6 5]\n",
" [ 6 180 1 6 57 1 4 4 8 14]\n",
" [ 1 3 111 3 3 55 2 25 0 3]\n",
" [ 49 4 3 122 13 1 17 4 1 2]\n",
" [ 7 42 3 24 276 6 7 1 11 15]\n",
" [ 1 0 48 1 8 121 0 24 0 3]\n",
" [ 19 2 1 12 9 0 127 1 0 0]\n",
" [ 0 1 21 2 1 21 1 134 0 1]\n",
" [ 4 14 1 3 10 0 0 0 126 1]\n",
" [ 2 6 2 4 21 1 2 3 0 81]]\n",
"val Loss: 0.7977 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[107 5 1 19 3 0 17 2 1 2]\n",
" [ 8 112 0 2 25 1 8 4 5 2]\n",
" [ 2 0 71 0 0 31 0 17 0 2]\n",
" [ 19 7 1 91 6 0 4 0 0 1]\n",
" [ 6 17 1 10 178 2 3 1 4 11]\n",
" [ 0 0 25 0 1 78 0 17 0 2]\n",
" [ 8 2 0 6 2 0 83 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 2 1 3 6 0 1 0 0 55]]\n",
"\n",
"Epoch 52/119\n",
"----------\n",
"train Loss: 0.9442 Acc: 0.6538\n",
"Confusion Matrix:\n",
"[[158 10 4 39 21 0 25 0 4 2]\n",
" [ 10 171 1 7 71 2 7 0 6 6]\n",
" [ 0 5 122 1 4 48 1 24 0 1]\n",
" [ 27 6 1 129 33 1 13 2 2 2]\n",
" [ 10 53 2 17 263 5 10 0 9 23]\n",
" [ 2 4 32 3 6 133 0 23 0 3]\n",
" [ 22 2 2 10 11 0 122 1 0 1]\n",
" [ 0 0 16 0 0 24 2 139 0 1]\n",
" [ 4 16 0 3 11 0 2 0 121 2]\n",
" [ 2 7 0 3 26 0 2 2 1 79]]\n",
"val Loss: 0.8069 Acc: 0.7357\n",
"Confusion Matrix:\n",
"[[113 6 1 20 1 0 13 1 0 2]\n",
" [ 8 110 0 1 30 1 7 4 3 3]\n",
" [ 2 0 68 0 0 36 0 15 0 2]\n",
" [ 18 2 1 91 11 0 4 0 0 2]\n",
" [ 6 12 1 10 185 2 3 1 1 12]\n",
" [ 0 0 24 1 1 87 0 8 0 2]\n",
" [ 10 2 0 7 3 0 79 0 0 0]\n",
" [ 0 0 9 0 1 14 0 85 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 2 1 1 3 6 0 0 0 0 59]]\n",
"\n",
"Epoch 53/119\n",
"----------\n",
"train Loss: 0.9423 Acc: 0.6770\n",
"Confusion Matrix:\n",
"[[166 8 0 43 15 0 22 1 4 4]\n",
" [ 11 184 1 6 50 2 8 3 8 8]\n",
" [ 2 3 124 1 2 48 1 23 0 2]\n",
" [ 30 8 0 133 19 2 23 0 1 0]\n",
" [ 13 46 7 14 278 5 5 0 6 18]\n",
" [ 0 7 33 3 7 128 1 24 0 3]\n",
" [ 17 5 0 12 5 0 130 0 1 1]\n",
" [ 0 1 19 0 1 19 0 141 0 1]\n",
" [ 4 12 1 2 11 2 0 0 127 0]\n",
" [ 0 5 1 6 27 1 2 3 0 77]]\n",
"val Loss: 0.8126 Acc: 0.7349\n",
"Confusion Matrix:\n",
"[[110 5 1 18 3 0 12 3 3 2]\n",
" [ 5 113 0 0 29 1 7 4 6 2]\n",
" [ 1 1 64 0 0 34 0 21 0 2]\n",
" [ 21 3 1 83 15 0 3 1 0 2]\n",
" [ 4 12 0 8 190 3 3 1 2 10]\n",
" [ 0 0 20 0 1 83 0 17 0 2]\n",
" [ 8 3 0 7 4 0 79 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 2 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 54/119\n",
"----------\n",
"train Loss: 0.9547 Acc: 0.6697\n",
"Confusion Matrix:\n",
"[[167 11 4 34 13 0 22 2 7 3]\n",
" [ 13 184 1 6 53 4 5 4 6 5]\n",
" [ 5 0 130 3 5 41 3 17 0 2]\n",
" [ 40 5 2 120 28 1 14 2 1 3]\n",
" [ 11 39 5 16 279 7 7 4 6 18]\n",
" [ 3 7 37 1 9 122 0 27 0 0]\n",
" [ 21 6 1 10 4 0 128 1 0 0]\n",
" [ 0 3 15 1 0 23 3 137 0 0]\n",
" [ 1 10 0 3 15 0 1 0 128 1]\n",
" [ 1 4 5 5 26 0 2 0 2 77]]\n",
"val Loss: 0.8087 Acc: 0.7235\n",
"Confusion Matrix:\n",
"[[112 4 1 20 1 0 14 1 2 2]\n",
" [ 12 104 0 1 28 1 8 4 7 2]\n",
" [ 2 0 73 0 0 31 0 15 0 2]\n",
" [ 27 4 1 85 8 0 4 0 0 0]\n",
" [ 7 13 1 10 180 2 3 1 6 10]\n",
" [ 1 1 29 1 1 77 0 11 0 2]\n",
" [ 10 1 0 6 2 0 82 0 0 0]\n",
" [ 0 0 7 0 1 9 0 92 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 5 1 1 2 7 0 1 0 0 55]]\n",
"\n",
"Epoch 55/119\n",
"----------\n",
"train Loss: 0.9523 Acc: 0.6661\n",
"Confusion Matrix:\n",
"[[169 7 2 34 20 1 25 3 2 0]\n",
" [ 13 171 0 9 60 3 9 3 6 7]\n",
" [ 3 3 124 1 4 42 0 26 0 3]\n",
" [ 45 4 1 122 22 1 16 1 1 3]\n",
" [ 11 40 7 19 271 2 9 4 9 20]\n",
" [ 0 6 42 1 9 124 2 22 0 0]\n",
" [ 12 8 0 10 6 0 133 2 0 0]\n",
" [ 1 2 19 0 1 15 3 141 0 0]\n",
" [ 4 10 1 3 10 0 2 0 128 1]\n",
" [ 3 9 2 3 19 0 1 2 2 81]]\n",
"val Loss: 0.8086 Acc: 0.7235\n",
"Confusion Matrix:\n",
"[[118 5 1 15 2 0 12 1 1 2]\n",
" [ 9 119 0 0 20 1 7 4 5 2]\n",
" [ 2 0 73 0 0 31 0 15 0 2]\n",
" [ 31 8 1 80 6 0 3 0 0 0]\n",
" [ 6 20 1 10 175 1 3 1 5 11]\n",
" [ 1 0 30 1 1 75 0 13 0 2]\n",
" [ 14 1 0 6 3 0 77 0 0 0]\n",
" [ 0 0 9 0 1 10 0 89 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 1 2 8 0 0 0 0 55]]\n",
"\n",
"Epoch 56/119\n",
"----------\n",
"train Loss: 0.9493 Acc: 0.6679\n",
"Confusion Matrix:\n",
"[[173 6 2 30 21 0 21 2 4 4]\n",
" [ 12 180 2 10 53 2 3 4 4 11]\n",
" [ 1 2 115 2 5 53 3 22 0 3]\n",
" [ 45 5 1 131 15 0 16 1 2 0]\n",
" [ 14 49 6 20 280 1 4 1 4 13]\n",
" [ 0 3 47 3 7 124 0 22 0 0]\n",
" [ 16 6 0 15 1 0 131 0 1 1]\n",
" [ 0 0 17 0 0 27 0 138 0 0]\n",
" [ 6 13 1 4 13 0 1 1 119 1]\n",
" [ 6 5 1 1 26 1 2 3 0 77]]\n",
"val Loss: 0.8053 Acc: 0.7418\n",
"Confusion Matrix:\n",
"[[110 5 2 23 1 0 11 1 2 2]\n",
" [ 6 116 0 0 26 1 7 4 5 2]\n",
" [ 2 0 65 0 1 35 0 18 0 2]\n",
" [ 16 4 0 96 9 0 2 1 0 1]\n",
" [ 5 17 1 10 181 1 3 1 4 10]\n",
" [ 0 0 21 1 1 84 0 14 0 2]\n",
" [ 8 3 0 7 2 0 81 0 0 0]\n",
" [ 0 0 4 0 1 11 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 3 6 0 0 0 0 59]]\n",
"\n",
"Epoch 57/119\n",
"----------\n",
"train Loss: 0.9483 Acc: 0.6761\n",
"Confusion Matrix:\n",
"[[169 11 3 28 16 1 26 3 5 1]\n",
" [ 13 185 1 8 49 1 4 5 6 9]\n",
" [ 2 0 129 2 6 43 1 16 0 7]\n",
" [ 37 3 1 126 25 1 19 4 0 0]\n",
" [ 10 51 3 14 288 4 5 2 8 7]\n",
" [ 3 6 37 2 9 121 1 23 1 3]\n",
" [ 23 2 1 16 8 1 118 0 1 1]\n",
" [ 0 0 17 0 0 20 0 144 0 1]\n",
" [ 4 6 0 4 11 0 1 1 131 1]\n",
" [ 4 9 2 3 26 1 0 1 1 75]]\n",
"val Loss: 0.8015 Acc: 0.7257\n",
"Confusion Matrix:\n",
"[[116 5 1 16 2 0 13 1 1 2]\n",
" [ 10 111 0 0 28 0 7 4 4 3]\n",
" [ 2 0 73 0 0 28 0 18 0 2]\n",
" [ 28 2 0 84 9 0 4 0 0 2]\n",
" [ 8 14 0 10 182 1 3 1 3 11]\n",
" [ 2 1 35 1 1 66 0 14 0 3]\n",
" [ 11 1 0 5 2 0 82 0 0 0]\n",
" [ 0 0 6 0 1 8 0 93 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 1 1 3 6 0 1 0 0 57]]\n",
"\n",
"Epoch 58/119\n",
"----------\n",
"train Loss: 0.9406 Acc: 0.6793\n",
"Confusion Matrix:\n",
"[[168 12 4 33 21 1 18 0 2 4]\n",
" [ 13 180 2 15 51 5 4 4 4 3]\n",
" [ 1 2 129 1 10 38 3 19 0 3]\n",
" [ 35 6 1 132 21 1 16 1 2 1]\n",
" [ 14 48 5 23 271 3 6 1 8 13]\n",
" [ 3 2 36 1 2 139 0 21 0 2]\n",
" [ 17 2 0 9 7 0 133 1 2 0]\n",
" [ 0 1 15 1 0 25 2 136 0 2]\n",
" [ 2 16 2 3 8 0 2 1 125 0]\n",
" [ 1 8 3 1 26 1 1 1 0 80]]\n",
"val Loss: 0.8009 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[114 5 1 20 2 0 10 1 2 2]\n",
" [ 8 111 0 0 27 1 7 4 7 2]\n",
" [ 1 1 71 0 0 36 0 12 0 2]\n",
" [ 21 6 1 89 7 0 4 0 0 1]\n",
" [ 6 15 0 10 183 1 3 1 4 10]\n",
" [ 0 1 26 1 2 83 0 8 0 2]\n",
" [ 9 4 0 7 2 0 79 0 0 0]\n",
" [ 0 0 10 0 1 15 0 83 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 1 1 2 7 0 0 0 0 58]]\n",
"\n",
"Epoch 59/119\n",
"----------\n",
"train Loss: 0.9405 Acc: 0.6761\n",
"Confusion Matrix:\n",
"[[170 10 0 32 18 1 19 1 5 7]\n",
" [ 11 183 0 7 54 1 4 6 6 9]\n",
" [ 0 2 123 3 6 51 1 15 1 4]\n",
" [ 37 8 2 127 19 3 18 1 0 1]\n",
" [ 7 47 7 13 281 3 6 2 10 16]\n",
" [ 1 3 42 3 8 124 1 20 2 2]\n",
" [ 18 2 2 16 6 0 127 0 0 0]\n",
" [ 0 0 15 1 1 11 0 153 0 1]\n",
" [ 3 10 0 3 22 0 0 0 120 1]\n",
" [ 3 8 0 2 24 1 2 3 1 78]]\n",
"val Loss: 0.7947 Acc: 0.7357\n",
"Confusion Matrix:\n",
"[[114 5 1 20 1 0 12 1 1 2]\n",
" [ 9 114 0 1 26 1 7 4 3 2]\n",
" [ 2 0 69 0 0 35 0 15 0 2]\n",
" [ 21 7 1 89 6 0 4 0 0 1]\n",
" [ 7 15 0 10 184 2 3 1 1 10]\n",
" [ 1 0 24 1 1 82 0 12 0 2]\n",
" [ 12 1 0 5 2 0 81 0 0 0]\n",
" [ 0 0 5 0 1 12 0 91 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 3 1 3 7 0 1 0 0 53]]\n",
"\n",
"Epoch 60/119\n",
"----------\n",
"train Loss: 0.9575 Acc: 0.6597\n",
"Confusion Matrix:\n",
"[[163 13 2 33 15 0 22 0 9 6]\n",
" [ 13 186 2 6 50 2 6 2 10 4]\n",
" [ 2 3 119 1 9 52 1 19 0 0]\n",
" [ 32 10 3 122 26 3 14 2 2 2]\n",
" [ 4 38 3 16 290 5 8 2 8 18]\n",
" [ 2 4 53 4 7 112 0 21 0 3]\n",
" [ 24 4 1 16 7 0 118 1 0 0]\n",
" [ 0 1 22 0 1 13 0 145 0 0]\n",
" [ 3 12 0 5 16 0 3 1 118 1]\n",
" [ 1 4 1 5 28 3 0 3 0 77]]\n",
"val Loss: 0.8034 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[107 5 1 17 3 0 19 2 1 2]\n",
" [ 9 113 0 0 26 1 8 4 4 2]\n",
" [ 2 0 76 0 0 27 0 16 0 2]\n",
" [ 19 5 1 90 7 0 6 0 0 1]\n",
" [ 6 17 1 10 180 2 3 1 2 11]\n",
" [ 0 0 33 1 1 72 0 14 0 2]\n",
" [ 7 1 0 4 2 0 87 0 0 0]\n",
" [ 0 0 7 0 1 9 0 92 0 0]\n",
" [ 0 2 0 1 5 0 1 0 86 0]\n",
" [ 3 2 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 61/119\n",
"----------\n",
"train Loss: 0.9419 Acc: 0.6756\n",
"Confusion Matrix:\n",
"[[174 15 2 26 13 0 20 3 3 7]\n",
" [ 12 172 2 6 67 3 5 2 7 5]\n",
" [ 1 1 117 0 10 48 2 24 0 3]\n",
" [ 40 8 1 134 17 3 10 1 0 2]\n",
" [ 8 38 4 20 301 2 4 0 2 13]\n",
" [ 1 4 38 2 10 121 1 26 0 3]\n",
" [ 17 3 2 12 10 1 125 0 0 1]\n",
" [ 0 1 14 0 0 24 3 137 0 3]\n",
" [ 4 13 0 3 12 0 0 1 125 1]\n",
" [ 0 9 1 2 25 1 2 2 1 79]]\n",
"val Loss: 0.8099 Acc: 0.7273\n",
"Confusion Matrix:\n",
"[[109 6 1 19 3 0 14 1 2 2]\n",
" [ 5 117 0 0 27 1 7 4 4 2]\n",
" [ 2 1 63 0 1 33 0 21 0 2]\n",
" [ 24 6 0 87 7 0 4 0 0 1]\n",
" [ 6 18 0 9 180 1 3 1 6 9]\n",
" [ 2 1 21 1 1 78 0 16 0 3]\n",
" [ 10 2 0 6 3 0 80 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 3 1 2 8 0 1 0 0 55]]\n",
"\n",
"Epoch 62/119\n",
"----------\n",
"train Loss: 0.9408 Acc: 0.6788\n",
"Confusion Matrix:\n",
"[[174 11 2 30 15 0 21 0 6 4]\n",
" [ 10 190 3 7 51 4 4 1 6 5]\n",
" [ 2 4 121 0 4 42 3 27 0 3]\n",
" [ 44 5 4 123 21 0 17 1 1 0]\n",
" [ 13 40 7 15 284 3 5 0 9 16]\n",
" [ 1 3 37 1 8 129 1 24 1 1]\n",
" [ 12 3 2 18 5 0 131 0 0 0]\n",
" [ 0 0 12 0 2 28 2 136 0 2]\n",
" [ 1 8 0 8 15 0 2 0 123 2]\n",
" [ 5 3 1 1 25 2 1 1 2 81]]\n",
"val Loss: 0.8038 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[115 3 2 20 1 0 12 2 0 2]\n",
" [ 11 102 0 1 33 1 7 4 4 4]\n",
" [ 2 0 74 0 0 29 0 16 0 2]\n",
" [ 24 2 1 88 9 0 3 0 0 2]\n",
" [ 6 8 1 10 185 3 3 1 3 13]\n",
" [ 0 0 27 0 1 80 0 13 0 2]\n",
" [ 11 1 1 7 2 0 79 0 0 0]\n",
" [ 0 0 7 0 1 10 0 91 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 3 6 0 0 0 0 59]]\n",
"\n",
"Epoch 63/119\n",
"----------\n",
"train Loss: 0.9575 Acc: 0.6747\n",
"Confusion Matrix:\n",
"[[173 11 2 32 18 2 19 1 3 2]\n",
" [ 5 178 0 14 59 3 4 3 3 12]\n",
" [ 1 2 127 1 4 41 3 20 1 6]\n",
" [ 34 9 0 120 30 2 17 1 3 0]\n",
" [ 12 43 7 18 280 5 2 2 4 19]\n",
" [ 0 7 33 2 7 128 3 26 0 0]\n",
" [ 27 6 1 10 5 0 120 0 1 1]\n",
" [ 0 2 11 0 1 19 0 149 0 0]\n",
" [ 2 20 2 4 7 0 0 0 124 0]\n",
" [ 0 2 0 5 24 3 0 3 1 84]]\n",
"val Loss: 0.8147 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[107 4 1 19 3 0 16 3 2 2]\n",
" [ 5 117 0 0 25 2 7 4 5 2]\n",
" [ 1 1 63 0 0 34 0 22 0 2]\n",
" [ 18 6 1 89 9 0 4 1 0 1]\n",
" [ 6 17 1 8 180 3 3 1 5 9]\n",
" [ 0 0 19 1 1 81 0 19 0 2]\n",
" [ 8 4 0 6 3 0 80 0 0 0]\n",
" [ 0 0 1 0 1 9 0 98 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 3 1 2 8 0 1 1 0 55]]\n",
"\n",
"Epoch 64/119\n",
"----------\n",
"train Loss: 0.9306 Acc: 0.6888\n",
"Confusion Matrix:\n",
"[[171 12 1 30 21 1 17 1 3 6]\n",
" [ 10 185 0 7 51 3 6 5 4 10]\n",
" [ 0 2 130 1 4 39 0 24 1 5]\n",
" [ 30 6 0 140 20 2 11 2 2 3]\n",
" [ 11 38 7 15 280 3 7 2 12 17]\n",
" [ 2 2 37 1 11 123 0 30 0 0]\n",
" [ 16 4 0 12 2 0 136 1 0 0]\n",
" [ 0 1 19 1 0 12 0 147 0 2]\n",
" [ 4 17 0 5 11 0 0 0 121 1]\n",
" [ 2 7 2 1 24 2 1 1 1 81]]\n",
"val Loss: 0.8025 Acc: 0.7280\n",
"Confusion Matrix:\n",
"[[106 6 1 19 3 0 16 2 2 2]\n",
" [ 6 119 0 2 21 1 7 4 5 2]\n",
" [ 2 0 69 0 0 25 0 25 0 2]\n",
" [ 20 7 1 88 7 0 4 1 0 1]\n",
" [ 5 18 1 10 178 3 3 1 4 10]\n",
" [ 0 0 28 1 1 68 0 23 0 2]\n",
" [ 7 3 0 6 2 0 83 0 0 0]\n",
" [ 0 0 3 0 1 7 0 98 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 7 0 0 0 0 58]]\n",
"\n",
"Epoch 65/119\n",
"----------\n",
"train Loss: 0.9677 Acc: 0.6624\n",
"Confusion Matrix:\n",
"[[161 13 4 30 13 0 30 1 7 4]\n",
" [ 10 174 1 9 64 5 6 3 1 8]\n",
" [ 2 0 127 1 2 44 2 21 0 7]\n",
" [ 42 12 2 122 20 2 13 0 0 3]\n",
" [ 12 48 5 15 276 8 5 0 5 18]\n",
" [ 0 3 42 1 4 130 1 24 0 1]\n",
" [ 22 8 1 11 7 0 118 1 1 2]\n",
" [ 0 3 13 0 0 23 1 141 0 1]\n",
" [ 2 11 0 4 12 0 0 0 129 1]\n",
" [ 1 6 1 5 26 2 2 0 1 78]]\n",
"val Loss: 0.7961 Acc: 0.7418\n",
"Confusion Matrix:\n",
"[[113 6 1 18 2 0 13 1 1 2]\n",
" [ 8 110 0 2 28 1 7 4 5 2]\n",
" [ 2 0 70 0 1 32 0 16 0 2]\n",
" [ 17 3 1 93 10 0 4 0 0 1]\n",
" [ 5 13 0 10 187 2 3 1 3 9]\n",
" [ 0 1 25 1 1 84 0 9 0 2]\n",
" [ 8 3 0 7 2 0 81 0 0 0]\n",
" [ 0 0 6 0 1 11 0 91 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 2 1 3 6 0 1 0 0 56]]\n",
"\n",
"Epoch 66/119\n",
"----------\n",
"train Loss: 0.9511 Acc: 0.6642\n",
"Confusion Matrix:\n",
"[[169 9 2 38 14 1 23 0 3 4]\n",
" [ 12 177 0 10 59 4 4 3 6 6]\n",
" [ 0 6 121 1 7 47 3 16 1 4]\n",
" [ 32 4 1 132 27 2 12 2 4 0]\n",
" [ 13 43 2 17 278 8 1 2 7 21]\n",
" [ 1 1 51 2 9 120 0 19 0 3]\n",
" [ 15 6 1 15 11 1 121 1 0 0]\n",
" [ 0 1 12 0 1 13 3 149 0 3]\n",
" [ 3 11 0 5 19 0 0 0 121 0]\n",
" [ 4 5 3 3 30 0 0 2 3 72]]\n",
"val Loss: 0.8078 Acc: 0.7380\n",
"Confusion Matrix:\n",
"[[113 6 1 18 2 0 14 1 0 2]\n",
" [ 8 111 0 1 27 1 7 4 3 5]\n",
" [ 2 0 67 0 1 35 0 16 0 2]\n",
" [ 24 3 1 86 10 0 4 0 0 1]\n",
" [ 6 13 0 9 188 1 3 1 1 11]\n",
" [ 0 0 22 1 1 84 0 12 0 3]\n",
" [ 9 2 0 6 2 0 82 0 0 0]\n",
" [ 0 0 7 0 1 9 0 91 0 1]\n",
" [ 0 2 0 2 6 0 0 0 85 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 67/119\n",
"----------\n",
"train Loss: 0.9603 Acc: 0.6656\n",
"Confusion Matrix:\n",
"[[170 8 3 34 20 0 21 0 5 2]\n",
" [ 5 179 2 6 63 4 9 1 9 3]\n",
" [ 0 1 111 3 10 55 1 21 0 4]\n",
" [ 33 8 1 135 18 4 13 0 1 3]\n",
" [ 14 57 9 18 261 4 6 1 7 15]\n",
" [ 2 4 39 4 5 127 0 20 0 5]\n",
" [ 20 6 1 11 6 0 126 0 0 1]\n",
" [ 0 1 15 1 0 15 1 147 0 2]\n",
" [ 6 9 1 4 13 0 0 1 125 0]\n",
" [ 0 12 0 2 22 2 0 1 1 82]]\n",
"val Loss: 0.8088 Acc: 0.7341\n",
"Confusion Matrix:\n",
"[[113 3 1 23 1 0 10 3 1 2]\n",
" [ 7 111 0 2 28 1 7 4 5 2]\n",
" [ 2 0 64 0 0 34 0 21 0 2]\n",
" [ 21 3 0 91 10 0 2 1 0 1]\n",
" [ 7 10 1 11 184 3 2 1 4 10]\n",
" [ 0 0 21 1 1 79 0 18 0 3]\n",
" [ 12 2 0 7 3 0 77 0 0 0]\n",
" [ 0 0 2 0 1 10 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 3 5 0 0 0 0 60]]\n",
"\n",
"Epoch 68/119\n",
"----------\n",
"train Loss: 0.9258 Acc: 0.6884\n",
"Confusion Matrix:\n",
"[[175 8 1 33 15 0 20 2 6 3]\n",
" [ 10 188 3 10 51 2 2 2 5 8]\n",
" [ 2 3 128 0 2 46 1 21 0 3]\n",
" [ 34 5 1 133 28 1 10 2 0 2]\n",
" [ 12 47 1 18 275 7 7 1 5 19]\n",
" [ 0 1 43 0 4 133 2 19 0 4]\n",
" [ 22 1 0 10 4 0 133 1 0 0]\n",
" [ 0 0 13 0 1 20 2 145 0 1]\n",
" [ 7 15 1 2 12 0 1 0 121 0]\n",
" [ 3 4 3 3 21 2 1 2 1 82]]\n",
"val Loss: 0.7989 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[111 6 1 19 2 0 14 1 1 2]\n",
" [ 7 113 0 0 31 0 7 4 3 2]\n",
" [ 1 1 73 0 0 32 0 14 0 2]\n",
" [ 23 4 1 87 9 0 4 0 0 1]\n",
" [ 6 14 0 10 187 2 3 1 1 9]\n",
" [ 0 0 31 0 1 75 0 14 0 2]\n",
" [ 9 3 0 7 2 0 80 0 0 0]\n",
" [ 0 0 7 0 1 11 0 90 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 2 1 2 8 0 1 0 0 55]]\n",
"\n",
"Epoch 69/119\n",
"----------\n",
"train Loss: 0.9609 Acc: 0.6811\n",
"Confusion Matrix:\n",
"[[174 14 2 26 22 1 17 0 5 2]\n",
" [ 18 186 1 3 50 4 3 1 7 8]\n",
" [ 0 2 125 0 7 43 2 23 0 4]\n",
" [ 39 2 2 125 24 5 15 1 2 1]\n",
" [ 7 43 3 17 286 5 4 3 7 17]\n",
" [ 1 2 30 1 10 137 2 18 0 5]\n",
" [ 21 1 1 10 6 1 129 0 1 1]\n",
" [ 0 1 25 2 0 15 2 136 0 1]\n",
" [ 4 15 2 4 12 0 0 0 122 0]\n",
" [ 3 6 0 4 27 1 1 1 2 77]]\n",
"val Loss: 0.8130 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[106 3 1 22 2 0 12 3 6 2]\n",
" [ 5 116 0 2 23 1 7 4 7 2]\n",
" [ 2 0 65 0 0 32 0 22 0 2]\n",
" [ 17 7 0 94 7 0 2 1 0 1]\n",
" [ 6 16 1 11 178 3 2 1 5 10]\n",
" [ 0 0 20 1 1 81 0 18 0 2]\n",
" [ 12 1 0 7 4 0 77 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 3 0 2 3 0 0 0 87 0]\n",
" [ 3 2 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 70/119\n",
"----------\n",
"train Loss: 0.9557 Acc: 0.6538\n",
"Confusion Matrix:\n",
"[[165 13 2 32 21 2 17 1 7 3]\n",
" [ 12 180 1 9 53 7 3 5 6 5]\n",
" [ 2 3 120 0 6 47 2 23 0 3]\n",
" [ 35 9 3 113 24 4 21 1 3 3]\n",
" [ 12 61 5 19 258 4 6 0 6 21]\n",
" [ 1 4 29 1 8 135 1 23 0 4]\n",
" [ 13 3 1 17 4 0 132 0 0 1]\n",
" [ 0 0 20 0 2 20 2 135 0 3]\n",
" [ 4 13 0 9 8 0 0 0 124 1]\n",
" [ 1 7 0 1 31 2 2 2 1 75]]\n",
"val Loss: 0.8081 Acc: 0.7364\n",
"Confusion Matrix:\n",
"[[103 6 1 22 2 0 17 2 2 2]\n",
" [ 5 114 0 3 26 1 7 4 5 2]\n",
" [ 2 0 64 0 1 32 0 22 0 2]\n",
" [ 15 4 0 94 10 0 4 1 0 1]\n",
" [ 5 16 0 10 185 2 3 1 2 9]\n",
" [ 0 0 19 1 1 82 0 18 0 2]\n",
" [ 7 3 0 6 3 0 82 0 0 0]\n",
" [ 0 0 2 0 1 9 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 71/119\n",
"----------\n",
"train Loss: 0.9304 Acc: 0.6806\n",
"Confusion Matrix:\n",
"[[165 16 4 33 15 0 18 0 6 6]\n",
" [ 11 190 0 10 46 4 5 2 4 9]\n",
" [ 1 1 117 2 8 52 1 19 1 4]\n",
" [ 35 6 2 125 21 2 21 0 3 1]\n",
" [ 6 42 7 12 295 4 5 1 3 17]\n",
" [ 3 5 44 1 9 120 0 21 0 3]\n",
" [ 17 3 0 14 3 0 134 0 0 0]\n",
" [ 0 1 18 0 0 18 0 144 0 1]\n",
" [ 1 15 0 5 8 0 2 0 127 1]\n",
" [ 2 3 2 3 28 0 4 1 0 79]]\n",
"val Loss: 0.7971 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[111 6 1 18 2 0 14 1 2 2]\n",
" [ 7 117 0 1 27 0 7 3 3 2]\n",
" [ 2 0 70 0 1 33 0 12 1 4]\n",
" [ 20 6 0 91 7 0 4 0 0 1]\n",
" [ 7 14 0 10 184 2 3 1 2 10]\n",
" [ 2 1 24 1 2 78 0 13 0 2]\n",
" [ 9 4 0 6 3 0 79 0 0 0]\n",
" [ 0 0 6 0 1 12 0 89 0 1]\n",
" [ 0 3 0 2 5 0 0 0 85 0]\n",
" [ 2 3 1 4 7 0 1 0 0 54]]\n",
"\n",
"Epoch 72/119\n",
"----------\n",
"train Loss: 0.9624 Acc: 0.6638\n",
"Confusion Matrix:\n",
"[[182 8 2 27 11 1 24 1 4 3]\n",
" [ 11 182 3 6 61 2 4 2 3 7]\n",
" [ 4 3 110 2 6 51 3 21 0 6]\n",
" [ 39 1 4 128 21 0 17 2 2 2]\n",
" [ 8 51 4 21 270 5 4 2 10 17]\n",
" [ 3 2 28 2 13 133 2 19 1 3]\n",
" [ 22 5 1 13 5 0 124 0 0 1]\n",
" [ 1 1 22 1 2 21 2 131 1 0]\n",
" [ 6 10 2 2 11 0 0 1 125 2]\n",
" [ 2 14 3 2 24 0 1 1 1 74]]\n",
"val Loss: 0.8088 Acc: 0.7288\n",
"Confusion Matrix:\n",
"[[109 5 2 20 3 0 12 1 3 2]\n",
" [ 4 118 0 0 24 1 7 4 7 2]\n",
" [ 2 0 72 0 1 23 0 23 0 2]\n",
" [ 24 5 1 83 12 0 3 1 0 0]\n",
" [ 5 17 1 9 180 1 3 1 6 10]\n",
" [ 0 0 29 1 1 70 0 20 0 2]\n",
" [ 8 4 0 7 3 0 79 0 0 0]\n",
" [ 0 0 6 0 1 5 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 73/119\n",
"----------\n",
"train Loss: 0.9516 Acc: 0.6783\n",
"Confusion Matrix:\n",
"[[158 14 2 29 18 2 31 0 6 3]\n",
" [ 16 182 3 9 48 2 5 3 7 6]\n",
" [ 2 3 119 2 7 48 3 20 0 2]\n",
" [ 40 6 0 127 23 2 16 2 0 0]\n",
" [ 11 39 4 14 299 2 2 0 7 14]\n",
" [ 1 0 39 1 6 141 1 14 0 3]\n",
" [ 23 4 2 14 7 0 117 1 0 3]\n",
" [ 1 0 16 1 1 19 0 144 0 0]\n",
" [ 4 18 1 3 15 0 0 0 116 2]\n",
" [ 2 8 2 2 18 1 0 1 0 88]]\n",
"val Loss: 0.8120 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[107 5 1 21 2 0 16 1 2 2]\n",
" [ 8 112 0 1 26 1 7 4 6 2]\n",
" [ 2 0 65 0 1 31 0 22 0 2]\n",
" [ 17 5 1 93 8 0 4 0 0 1]\n",
" [ 7 14 0 10 181 2 3 1 6 9]\n",
" [ 0 0 20 1 1 80 0 19 0 2]\n",
" [ 8 2 0 7 2 0 82 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 3 1 2 7 0 1 0 0 55]]\n",
"\n",
"Epoch 74/119\n",
"----------\n",
"train Loss: 0.9486 Acc: 0.6679\n",
"Confusion Matrix:\n",
"[[168 8 0 28 21 2 22 4 6 4]\n",
" [ 15 180 4 4 57 1 5 2 3 10]\n",
" [ 2 0 118 3 7 51 2 18 0 5]\n",
" [ 43 4 0 137 15 1 12 1 2 1]\n",
" [ 8 47 7 16 277 4 4 2 8 19]\n",
" [ 1 6 43 0 6 120 1 27 0 2]\n",
" [ 21 4 2 13 5 1 125 0 0 0]\n",
" [ 0 1 28 1 0 18 2 131 0 1]\n",
" [ 2 11 0 1 11 0 1 0 132 1]\n",
" [ 2 5 2 4 24 1 2 2 0 80]]\n",
"val Loss: 0.8056 Acc: 0.7357\n",
"Confusion Matrix:\n",
"[[104 5 1 19 3 0 19 2 2 2]\n",
" [ 5 114 0 0 30 1 7 4 4 2]\n",
" [ 1 1 63 0 1 33 0 22 0 2]\n",
" [ 18 4 1 89 11 0 4 1 0 1]\n",
" [ 5 13 0 8 187 3 3 1 4 9]\n",
" [ 0 0 21 0 1 85 0 14 0 2]\n",
" [ 7 3 0 6 3 0 82 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 75/119\n",
"----------\n",
"train Loss: 0.9423 Acc: 0.6652\n",
"Confusion Matrix:\n",
"[[162 15 1 36 22 0 18 1 5 3]\n",
" [ 7 186 1 7 57 3 5 3 5 7]\n",
" [ 3 1 119 1 8 52 1 18 0 3]\n",
" [ 43 4 3 122 27 1 14 0 1 1]\n",
" [ 10 46 9 19 275 3 4 1 9 16]\n",
" [ 2 7 33 2 8 128 0 23 0 3]\n",
" [ 22 2 0 13 5 0 126 1 0 2]\n",
" [ 0 1 20 0 1 15 2 142 0 1]\n",
" [ 1 10 0 3 13 0 2 0 127 3]\n",
" [ 3 5 1 6 27 0 2 3 0 75]]\n",
"val Loss: 0.8015 Acc: 0.7387\n",
"Confusion Matrix:\n",
"[[110 4 1 19 3 0 14 2 2 2]\n",
" [ 8 112 0 1 26 1 7 4 6 2]\n",
" [ 2 0 65 0 0 33 0 21 0 2]\n",
" [ 17 3 1 92 11 0 3 1 0 1]\n",
" [ 5 13 1 10 182 3 3 1 5 10]\n",
" [ 0 0 19 0 1 85 0 16 0 2]\n",
" [ 8 3 0 6 3 0 81 0 0 0]\n",
" [ 0 0 1 0 1 10 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 2 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 76/119\n",
"----------\n",
"train Loss: 0.9660 Acc: 0.6611\n",
"Confusion Matrix:\n",
"[[168 6 1 33 22 2 22 2 5 2]\n",
" [ 12 179 2 8 53 2 5 5 3 12]\n",
" [ 2 3 105 3 5 57 3 20 0 8]\n",
" [ 53 7 2 117 19 1 14 1 1 1]\n",
" [ 9 46 8 20 279 4 8 0 3 15]\n",
" [ 2 2 39 0 10 126 0 25 0 2]\n",
" [ 13 8 1 11 8 0 129 1 0 0]\n",
" [ 1 2 14 0 1 14 1 146 1 2]\n",
" [ 2 14 0 4 13 0 0 0 126 0]\n",
" [ 2 4 4 5 26 0 0 1 2 78]]\n",
"val Loss: 0.8103 Acc: 0.7372\n",
"Confusion Matrix:\n",
"[[109 4 1 20 1 0 17 2 1 2]\n",
" [ 10 108 0 2 28 1 8 4 4 2]\n",
" [ 2 0 67 0 0 35 0 17 0 2]\n",
" [ 20 2 0 92 10 0 4 0 0 1]\n",
" [ 7 12 1 10 184 3 3 1 1 11]\n",
" [ 0 0 19 1 1 87 0 13 0 2]\n",
" [ 9 0 0 6 2 0 84 0 0 0]\n",
" [ 0 0 5 0 1 9 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 1 1 3 8 0 1 0 0 54]]\n",
"\n",
"Epoch 77/119\n",
"----------\n",
"train Loss: 0.9412 Acc: 0.6683\n",
"Confusion Matrix:\n",
"[[159 13 3 33 21 2 20 0 6 6]\n",
" [ 19 178 0 5 51 5 5 5 6 7]\n",
" [ 3 0 116 0 7 53 1 23 0 3]\n",
" [ 42 2 2 130 19 0 15 3 1 2]\n",
" [ 7 44 4 14 280 5 13 4 6 15]\n",
" [ 0 4 45 2 7 127 1 15 0 5]\n",
" [ 14 2 3 8 6 0 136 0 0 2]\n",
" [ 0 2 10 0 0 18 1 150 1 0]\n",
" [ 5 12 1 2 14 1 1 0 122 1]\n",
" [ 3 11 2 1 32 0 0 2 0 71]]\n",
"val Loss: 0.8143 Acc: 0.7380\n",
"Confusion Matrix:\n",
"[[111 7 1 20 3 0 9 1 3 2]\n",
" [ 5 120 0 0 23 1 7 4 5 2]\n",
" [ 1 1 67 0 0 37 0 15 0 2]\n",
" [ 22 6 1 87 10 0 2 0 0 1]\n",
" [ 5 17 0 8 185 1 3 1 3 10]\n",
" [ 0 1 21 1 1 89 0 8 0 2]\n",
" [ 11 5 0 7 4 0 74 0 0 0]\n",
" [ 0 0 6 0 1 13 0 89 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 7 0 0 0 0 58]]\n",
"\n",
"Epoch 78/119\n",
"----------\n",
"train Loss: 0.9532 Acc: 0.6642\n",
"Confusion Matrix:\n",
"[[173 9 2 30 18 2 18 0 5 6]\n",
" [ 12 179 1 7 56 6 3 3 5 9]\n",
" [ 2 4 109 2 11 52 0 22 0 4]\n",
" [ 45 5 2 118 24 1 13 2 5 1]\n",
" [ 9 41 1 13 282 4 9 2 9 22]\n",
" [ 1 5 39 1 7 127 0 24 0 2]\n",
" [ 25 1 4 12 4 0 124 0 0 1]\n",
" [ 1 3 15 0 0 18 0 145 0 0]\n",
" [ 5 10 1 3 8 0 2 1 128 1]\n",
" [ 5 4 4 3 29 0 1 0 1 75]]\n",
"val Loss: 0.8055 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[110 7 1 19 2 0 13 1 2 2]\n",
" [ 5 124 0 0 19 1 7 4 5 2]\n",
" [ 1 1 70 0 0 33 0 16 0 2]\n",
" [ 26 8 1 84 6 0 3 0 0 1]\n",
" [ 4 24 1 9 176 1 3 1 6 8]\n",
" [ 0 1 23 1 1 85 0 10 0 2]\n",
" [ 9 5 0 7 2 0 78 0 0 0]\n",
" [ 0 0 5 0 1 11 0 92 0 0]\n",
" [ 0 3 0 2 3 0 0 0 87 0]\n",
" [ 1 7 1 2 8 0 1 0 0 52]]\n",
"\n",
"Epoch 79/119\n",
"----------\n",
"train Loss: 0.9241 Acc: 0.6774\n",
"Confusion Matrix:\n",
"[[174 12 1 30 18 2 20 1 2 3]\n",
" [ 12 172 1 6 63 5 5 2 7 8]\n",
" [ 6 1 127 1 7 39 2 19 0 4]\n",
" [ 47 7 0 127 22 1 11 0 1 0]\n",
" [ 11 45 5 21 276 5 5 0 11 13]\n",
" [ 0 2 33 2 7 138 1 20 1 2]\n",
" [ 19 4 4 10 4 0 125 3 1 1]\n",
" [ 0 0 11 1 0 11 0 156 0 3]\n",
" [ 6 11 1 4 14 0 0 0 123 0]\n",
" [ 3 9 1 2 33 0 2 0 1 71]]\n",
"val Loss: 0.8019 Acc: 0.7433\n",
"Confusion Matrix:\n",
"[[112 4 1 17 3 0 15 2 1 2]\n",
" [ 6 115 0 1 28 1 7 4 3 2]\n",
" [ 2 0 68 0 0 31 0 20 0 2]\n",
" [ 20 6 1 88 8 0 4 1 0 1]\n",
" [ 5 13 1 9 188 2 3 1 1 10]\n",
" [ 0 0 22 0 1 84 0 14 0 2]\n",
" [ 8 3 0 6 2 0 82 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 2 1 2 8 0 1 0 0 55]]\n",
"\n",
"Epoch 80/119\n",
"----------\n",
"train Loss: 0.9348 Acc: 0.6742\n",
"Confusion Matrix:\n",
"[[172 11 1 32 15 1 19 3 5 4]\n",
" [ 12 175 0 6 57 4 4 5 6 12]\n",
" [ 2 5 123 1 5 46 1 19 0 4]\n",
" [ 41 5 4 127 25 1 11 0 1 1]\n",
" [ 9 38 3 16 285 7 6 1 8 19]\n",
" [ 0 3 36 2 10 130 2 22 0 1]\n",
" [ 21 3 1 11 3 2 128 1 0 1]\n",
" [ 0 1 16 0 1 21 2 139 0 2]\n",
" [ 5 11 1 4 9 1 2 0 125 1]\n",
" [ 1 6 6 3 24 1 2 1 0 78]]\n",
"val Loss: 0.8033 Acc: 0.7387\n",
"Confusion Matrix:\n",
"[[110 8 1 17 3 0 13 1 2 2]\n",
" [ 6 118 0 0 25 1 7 4 3 3]\n",
" [ 1 1 67 0 1 36 0 15 0 2]\n",
" [ 22 5 1 88 8 0 4 0 0 1]\n",
" [ 5 15 0 9 186 1 3 1 3 10]\n",
" [ 1 1 21 0 1 85 0 11 0 3]\n",
" [ 9 5 0 6 3 0 78 0 0 0]\n",
" [ 0 1 5 0 1 10 0 91 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 3 1 2 6 0 0 0 0 58]]\n",
"\n",
"Epoch 81/119\n",
"----------\n",
"train Loss: 0.9402 Acc: 0.6770\n",
"Confusion Matrix:\n",
"[[169 12 1 42 8 0 21 1 3 6]\n",
" [ 11 185 0 8 58 2 4 2 5 6]\n",
" [ 2 1 116 4 5 53 2 20 0 3]\n",
" [ 36 8 2 130 18 1 14 2 1 4]\n",
" [ 11 41 2 15 293 5 5 2 4 14]\n",
" [ 1 1 35 3 5 127 0 29 0 5]\n",
" [ 19 6 1 10 6 0 124 1 0 4]\n",
" [ 1 1 19 0 1 14 0 146 0 0]\n",
" [ 5 10 1 2 15 0 0 2 124 0]\n",
" [ 1 6 2 3 29 2 3 2 0 74]]\n",
"val Loss: 0.7967 Acc: 0.7288\n",
"Confusion Matrix:\n",
"[[106 5 1 19 3 0 17 3 1 2]\n",
" [ 6 118 0 1 25 1 7 4 3 2]\n",
" [ 2 0 67 0 0 31 0 21 0 2]\n",
" [ 21 6 1 87 8 0 4 1 0 1]\n",
" [ 4 16 1 9 184 2 3 2 1 11]\n",
" [ 0 0 31 0 1 71 0 18 0 2]\n",
" [ 8 2 0 6 2 0 83 0 0 0]\n",
" [ 0 0 4 0 1 7 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 1 1 1 9 0 2 0 0 55]]\n",
"\n",
"Epoch 82/119\n",
"----------\n",
"train Loss: 0.9472 Acc: 0.6742\n",
"Confusion Matrix:\n",
"[[161 12 1 34 23 1 19 1 4 7]\n",
" [ 15 171 0 9 54 5 8 4 6 9]\n",
" [ 5 1 127 2 4 45 3 15 0 4]\n",
" [ 35 11 2 132 14 1 17 1 1 2]\n",
" [ 15 36 7 16 285 3 4 2 9 15]\n",
" [ 2 2 36 0 10 130 0 25 0 1]\n",
" [ 20 8 0 12 4 0 127 0 0 0]\n",
" [ 1 1 20 0 1 16 1 140 0 2]\n",
" [ 4 9 3 3 12 1 1 0 126 0]\n",
" [ 1 8 0 2 25 2 1 0 0 83]]\n",
"val Loss: 0.8131 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[102 5 2 19 3 0 16 2 6 2]\n",
" [ 5 119 0 0 23 1 7 4 6 2]\n",
" [ 0 1 69 0 0 33 0 18 0 2]\n",
" [ 17 5 1 90 11 0 3 1 0 1]\n",
" [ 4 16 1 9 180 3 3 1 4 12]\n",
" [ 0 0 22 0 1 81 0 17 0 2]\n",
" [ 8 3 1 7 3 0 79 0 0 0]\n",
" [ 0 0 4 0 1 10 0 94 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 1 2 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 83/119\n",
"----------\n",
"train Loss: 0.9507 Acc: 0.6788\n",
"Confusion Matrix:\n",
"[[177 11 3 20 15 1 25 2 4 5]\n",
" [ 10 180 1 12 57 1 6 2 4 8]\n",
" [ 2 2 119 0 7 46 2 22 1 5]\n",
" [ 34 7 1 133 20 3 15 2 0 1]\n",
" [ 8 41 9 18 281 6 4 1 12 12]\n",
" [ 1 3 37 3 14 127 0 19 0 2]\n",
" [ 23 6 0 11 5 0 124 1 1 0]\n",
" [ 0 1 20 0 1 13 2 143 0 2]\n",
" [ 4 10 0 3 11 0 2 0 128 1]\n",
" [ 3 8 1 2 24 1 0 3 0 80]]\n",
"val Loss: 0.8040 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[111 4 1 20 2 0 14 2 1 2]\n",
" [ 6 115 0 2 25 1 7 4 5 2]\n",
" [ 2 0 72 0 0 29 0 18 0 2]\n",
" [ 23 4 1 85 10 0 4 1 0 1]\n",
" [ 5 17 1 10 181 2 3 1 3 10]\n",
" [ 0 0 30 0 1 74 0 16 0 2]\n",
" [ 9 2 0 7 3 0 80 0 0 0]\n",
" [ 0 0 4 0 1 8 0 96 0 0]\n",
" [ 0 2 0 2 3 0 0 0 88 0]\n",
" [ 2 1 1 2 7 0 1 0 0 58]]\n",
"\n",
"Epoch 84/119\n",
"----------\n",
"train Loss: 0.9431 Acc: 0.6720\n",
"Confusion Matrix:\n",
"[[166 21 1 25 13 0 26 1 7 3]\n",
" [ 14 179 2 5 51 2 8 2 7 11]\n",
" [ 1 2 120 1 12 43 3 20 0 4]\n",
" [ 47 4 0 125 22 1 10 1 2 4]\n",
" [ 14 38 8 13 285 7 5 2 4 16]\n",
" [ 0 2 48 2 6 124 0 22 0 2]\n",
" [ 23 3 0 9 10 1 123 1 1 0]\n",
" [ 0 0 17 0 0 13 1 148 1 2]\n",
" [ 1 11 0 4 13 0 2 0 127 1]\n",
" [ 2 6 2 3 18 2 2 2 5 80]]\n",
"val Loss: 0.8065 Acc: 0.7257\n",
"Confusion Matrix:\n",
"[[107 4 1 21 1 0 16 3 2 2]\n",
" [ 6 119 0 0 23 1 7 4 5 2]\n",
" [ 2 0 69 0 0 24 0 26 0 2]\n",
" [ 23 5 1 86 8 0 4 1 0 1]\n",
" [ 6 15 2 10 181 3 3 1 4 8]\n",
" [ 0 0 30 1 1 65 0 24 0 2]\n",
" [ 8 2 0 6 2 0 83 0 0 0]\n",
" [ 0 0 2 0 1 7 0 99 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 3 3 2 7 0 1 0 0 54]]\n",
"\n",
"Epoch 85/119\n",
"----------\n",
"train Loss: 0.9456 Acc: 0.6765\n",
"Confusion Matrix:\n",
"[[175 15 0 31 12 1 23 2 2 2]\n",
" [ 16 172 0 6 61 5 3 3 4 11]\n",
" [ 1 2 119 1 4 52 4 16 0 7]\n",
" [ 41 4 4 124 19 3 19 1 0 1]\n",
" [ 9 41 10 19 283 4 7 2 5 12]\n",
" [ 0 2 35 3 8 136 1 21 0 0]\n",
" [ 20 9 2 16 7 1 115 0 1 0]\n",
" [ 0 0 14 0 1 16 0 150 0 1]\n",
" [ 3 12 0 3 9 0 2 0 129 1]\n",
" [ 2 5 1 1 23 2 2 1 1 84]]\n",
"val Loss: 0.8048 Acc: 0.7418\n",
"Confusion Matrix:\n",
"[[110 5 1 20 3 0 14 1 1 2]\n",
" [ 6 116 0 1 27 1 7 4 3 2]\n",
" [ 2 0 68 0 1 35 0 15 0 2]\n",
" [ 18 5 1 93 8 0 3 0 0 1]\n",
" [ 6 12 0 10 186 3 3 1 2 10]\n",
" [ 0 0 21 1 1 87 0 11 0 2]\n",
" [ 9 2 0 7 3 0 80 0 0 0]\n",
" [ 0 0 6 0 1 13 0 89 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 86/119\n",
"----------\n",
"train Loss: 0.9607 Acc: 0.6524\n",
"Confusion Matrix:\n",
"[[162 9 1 34 16 0 25 2 9 5]\n",
" [ 12 179 1 11 56 2 3 5 3 9]\n",
" [ 2 2 128 1 5 40 1 21 0 6]\n",
" [ 50 4 0 118 26 1 14 1 0 2]\n",
" [ 6 41 3 26 274 7 8 1 9 17]\n",
" [ 0 3 45 2 12 113 2 26 0 3]\n",
" [ 20 5 4 10 9 0 122 0 0 1]\n",
" [ 0 4 19 1 0 25 1 130 0 2]\n",
" [ 1 17 0 4 5 0 1 0 130 1]\n",
" [ 1 5 3 4 27 1 1 1 1 78]]\n",
"val Loss: 0.8097 Acc: 0.7341\n",
"Confusion Matrix:\n",
"[[108 3 1 21 1 0 17 3 1 2]\n",
" [ 10 114 0 1 23 1 7 4 5 2]\n",
" [ 2 0 72 0 0 23 0 24 0 2]\n",
" [ 20 5 1 92 6 0 4 1 0 0]\n",
" [ 6 16 1 10 182 2 3 1 3 9]\n",
" [ 1 0 27 1 1 71 0 20 0 2]\n",
" [ 9 1 0 5 2 0 84 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 4 3 1 4 6 0 1 0 0 53]]\n",
"\n",
"Epoch 87/119\n",
"----------\n",
"train Loss: 0.9429 Acc: 0.6843\n",
"Confusion Matrix:\n",
"[[178 11 2 25 16 1 25 0 2 3]\n",
" [ 13 179 2 7 53 3 2 6 3 13]\n",
" [ 4 3 123 1 7 39 2 24 1 2]\n",
" [ 26 10 1 137 24 0 14 3 1 0]\n",
" [ 17 42 6 11 281 4 6 2 11 12]\n",
" [ 1 4 34 0 15 124 3 23 0 2]\n",
" [ 21 2 0 13 2 0 131 1 0 1]\n",
" [ 0 0 14 2 1 18 2 144 0 1]\n",
" [ 5 13 2 2 13 0 1 0 123 0]\n",
" [ 2 8 0 1 26 0 0 1 0 84]]\n",
"val Loss: 0.8023 Acc: 0.7288\n",
"Confusion Matrix:\n",
"[[109 7 1 19 3 0 13 1 2 2]\n",
" [ 6 113 0 1 28 1 7 4 5 2]\n",
" [ 1 1 71 0 0 31 0 17 0 2]\n",
" [ 24 5 1 86 8 0 3 1 0 1]\n",
" [ 5 16 0 9 182 2 3 1 4 11]\n",
" [ 0 0 32 0 1 75 0 13 0 2]\n",
" [ 9 4 0 7 2 0 79 0 0 0]\n",
" [ 0 0 5 0 1 9 0 94 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 1 1 2 7 0 0 0 0 58]]\n",
"\n",
"Epoch 88/119\n",
"----------\n",
"train Loss: 0.9245 Acc: 0.6911\n",
"Confusion Matrix:\n",
"[[180 10 4 32 12 1 16 2 5 1]\n",
" [ 16 186 1 8 48 8 2 2 3 7]\n",
" [ 3 4 116 1 9 46 4 21 0 2]\n",
" [ 33 7 0 132 23 0 12 5 3 1]\n",
" [ 12 41 3 11 288 5 8 2 6 16]\n",
" [ 2 7 31 0 5 133 0 28 0 0]\n",
" [ 21 6 1 12 2 0 127 1 0 1]\n",
" [ 0 0 17 0 2 21 1 140 0 1]\n",
" [ 4 11 0 6 12 0 0 1 125 0]\n",
" [ 0 8 1 2 17 0 1 1 0 92]]\n",
"val Loss: 0.8056 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[112 4 1 20 1 0 14 2 1 2]\n",
" [ 8 112 0 1 26 1 8 4 4 3]\n",
" [ 2 0 67 0 0 33 0 19 0 2]\n",
" [ 23 5 1 87 8 0 4 0 0 1]\n",
" [ 7 16 0 10 183 2 3 1 1 10]\n",
" [ 1 0 23 1 1 78 0 17 0 2]\n",
" [ 10 1 0 6 2 0 82 0 0 0]\n",
" [ 0 0 5 0 1 7 0 96 0 0]\n",
" [ 0 2 0 1 5 0 1 0 86 0]\n",
" [ 2 2 1 3 6 0 1 0 0 57]]\n",
"\n",
"Epoch 89/119\n",
"----------\n",
"train Loss: 0.9532 Acc: 0.6679\n",
"Confusion Matrix:\n",
"[[170 11 1 33 12 1 24 0 5 6]\n",
" [ 9 192 1 5 50 2 4 2 6 10]\n",
" [ 1 2 119 2 6 48 0 25 0 3]\n",
" [ 34 7 2 129 29 1 11 1 0 2]\n",
" [ 12 48 6 21 272 1 5 2 10 15]\n",
" [ 3 4 43 2 9 125 3 17 0 0]\n",
" [ 24 5 0 16 7 0 118 1 0 0]\n",
" [ 0 1 24 0 1 14 0 140 0 2]\n",
" [ 2 17 0 4 13 0 0 0 123 0]\n",
" [ 1 6 1 2 30 0 0 1 1 80]]\n",
"val Loss: 0.8129 Acc: 0.7296\n",
"Confusion Matrix:\n",
"[[110 5 1 21 2 0 13 3 0 2]\n",
" [ 6 119 0 1 23 1 7 4 4 2]\n",
" [ 2 0 69 0 0 28 0 22 0 2]\n",
" [ 24 7 1 83 8 0 4 1 0 1]\n",
" [ 5 17 1 10 180 3 3 1 3 10]\n",
" [ 0 0 24 1 1 72 0 23 0 2]\n",
" [ 9 4 0 7 2 0 79 0 0 0]\n",
" [ 0 0 3 0 1 7 0 98 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 90/119\n",
"----------\n",
"train Loss: 0.9546 Acc: 0.6706\n",
"Confusion Matrix:\n",
"[[165 11 2 34 13 1 25 3 5 4]\n",
" [ 16 187 3 1 48 6 5 2 4 9]\n",
" [ 4 3 120 1 5 42 3 23 0 5]\n",
" [ 34 4 1 127 27 0 16 2 3 2]\n",
" [ 9 44 7 14 281 3 6 3 8 17]\n",
" [ 0 3 44 1 11 124 0 22 0 1]\n",
" [ 15 5 3 13 6 0 127 0 0 2]\n",
" [ 0 0 19 1 2 18 1 140 0 1]\n",
" [ 3 10 1 1 15 0 2 1 125 1]\n",
" [ 1 7 4 2 26 0 2 2 0 78]]\n",
"val Loss: 0.8057 Acc: 0.7319\n",
"Confusion Matrix:\n",
"[[106 5 1 21 2 0 15 3 2 2]\n",
" [ 6 115 0 1 25 2 7 4 5 2]\n",
" [ 2 0 66 0 0 31 0 22 0 2]\n",
" [ 19 6 1 89 8 0 4 1 0 1]\n",
" [ 5 15 1 10 179 3 3 1 6 10]\n",
" [ 0 0 20 0 1 81 0 19 0 2]\n",
" [ 8 3 0 6 2 0 82 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 3 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 91/119\n",
"----------\n",
"train Loss: 0.9191 Acc: 0.6988\n",
"Confusion Matrix:\n",
"[[183 10 1 23 14 0 21 0 7 4]\n",
" [ 12 186 2 6 47 3 7 4 7 7]\n",
" [ 3 2 134 3 7 33 0 20 0 4]\n",
" [ 37 5 2 128 23 2 15 2 1 1]\n",
" [ 13 46 4 15 272 4 4 1 12 21]\n",
" [ 1 4 31 2 6 136 0 25 0 1]\n",
" [ 17 7 0 7 6 1 132 1 0 0]\n",
" [ 0 1 19 0 1 17 1 142 0 1]\n",
" [ 3 5 0 3 10 0 0 0 137 1]\n",
" [ 4 5 2 2 21 0 0 2 0 86]]\n",
"val Loss: 0.8053 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[102 6 1 18 4 0 16 2 6 2]\n",
" [ 4 118 0 0 25 1 7 4 6 2]\n",
" [ 2 0 64 0 1 36 0 18 0 2]\n",
" [ 21 7 1 82 12 0 4 1 0 1]\n",
" [ 3 17 0 8 185 3 3 1 4 9]\n",
" [ 0 0 21 0 1 90 0 9 0 2]\n",
" [ 8 3 0 6 3 0 81 0 0 0]\n",
" [ 0 0 4 0 1 12 0 92 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 2 1 2 9 0 1 0 0 56]]\n",
"\n",
"Epoch 92/119\n",
"----------\n",
"train Loss: 0.9391 Acc: 0.6788\n",
"Confusion Matrix:\n",
"[[176 7 4 30 17 1 20 0 3 5]\n",
" [ 12 183 1 8 48 3 6 5 6 9]\n",
" [ 2 4 121 2 4 37 5 25 1 5]\n",
" [ 38 4 1 127 24 2 14 1 1 4]\n",
" [ 10 43 5 25 271 3 2 2 11 20]\n",
" [ 1 4 41 0 4 129 1 25 0 1]\n",
" [ 16 2 0 13 4 0 135 0 0 1]\n",
" [ 0 0 15 2 0 14 0 150 1 0]\n",
" [ 3 9 0 3 20 1 1 0 121 1]\n",
" [ 1 3 0 3 31 2 2 1 0 79]]\n",
"val Loss: 0.8064 Acc: 0.7273\n",
"Confusion Matrix:\n",
"[[112 3 2 20 2 0 13 2 1 2]\n",
" [ 10 107 0 2 29 1 7 4 5 2]\n",
" [ 2 0 69 0 0 33 0 17 0 2]\n",
" [ 18 4 1 95 7 0 3 0 0 1]\n",
" [ 6 13 1 12 180 2 3 1 5 10]\n",
" [ 0 0 29 1 1 76 0 14 0 2]\n",
" [ 11 1 1 7 2 0 79 0 0 0]\n",
" [ 0 0 7 0 1 8 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 1 1 2 8 0 1 0 0 55]]\n",
"\n",
"Epoch 93/119\n",
"----------\n",
"train Loss: 0.9606 Acc: 0.6633\n",
"Confusion Matrix:\n",
"[[165 16 0 32 16 0 19 4 8 3]\n",
" [ 11 179 1 9 62 1 6 1 5 6]\n",
" [ 4 1 123 1 4 45 3 22 0 3]\n",
" [ 45 5 1 116 26 1 16 2 0 4]\n",
" [ 8 42 5 21 286 4 10 1 4 11]\n",
" [ 3 1 34 0 11 131 0 24 0 2]\n",
" [ 22 3 2 14 6 0 124 0 0 0]\n",
" [ 0 1 20 2 1 15 3 140 0 0]\n",
" [ 1 18 2 5 13 0 2 0 118 0]\n",
" [ 3 8 0 3 27 1 3 1 0 76]]\n",
"val Loss: 0.8061 Acc: 0.7280\n",
"Confusion Matrix:\n",
"[[110 5 1 18 2 0 14 3 2 2]\n",
" [ 9 113 0 0 28 1 7 4 3 2]\n",
" [ 2 0 67 0 0 31 0 21 0 2]\n",
" [ 26 6 1 83 7 0 4 1 0 1]\n",
" [ 6 15 0 9 182 2 3 1 4 11]\n",
" [ 0 0 24 1 1 77 0 18 0 2]\n",
" [ 8 3 0 6 3 0 81 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 2 0 1 5 0 1 0 86 0]\n",
" [ 2 2 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 94/119\n",
"----------\n",
"train Loss: 0.9427 Acc: 0.6656\n",
"Confusion Matrix:\n",
"[[167 17 1 38 11 0 23 0 3 3]\n",
" [ 10 187 2 13 43 7 4 3 6 6]\n",
" [ 0 2 124 0 7 49 0 19 0 5]\n",
" [ 35 4 1 135 21 1 17 0 0 2]\n",
" [ 13 54 4 21 263 5 7 0 5 20]\n",
" [ 0 2 41 5 6 126 1 24 0 1]\n",
" [ 28 5 0 16 7 0 115 0 0 0]\n",
" [ 0 1 24 0 0 13 2 139 2 1]\n",
" [ 1 11 1 4 7 0 2 1 130 2]\n",
" [ 1 7 5 3 24 3 1 1 0 77]]\n",
"val Loss: 0.8029 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[110 4 1 20 1 0 14 3 2 2]\n",
" [ 8 114 0 2 23 1 7 4 6 2]\n",
" [ 2 0 76 0 0 26 0 17 0 2]\n",
" [ 20 7 1 89 6 0 4 1 0 1]\n",
" [ 6 14 1 10 181 3 3 1 4 10]\n",
" [ 0 0 28 0 1 74 0 19 0 1]\n",
" [ 9 1 1 6 2 0 82 0 0 0]\n",
" [ 0 0 6 0 1 9 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 2 2 7 0 1 0 0 54]]\n",
"\n",
"Epoch 95/119\n",
"----------\n",
"train Loss: 0.9478 Acc: 0.6720\n",
"Confusion Matrix:\n",
"[[175 11 1 33 14 0 21 1 1 6]\n",
" [ 12 175 3 6 56 4 10 1 5 9]\n",
" [ 3 1 118 2 5 51 2 21 0 3]\n",
" [ 42 6 1 123 22 2 13 2 4 1]\n",
" [ 17 38 3 20 281 5 7 3 5 13]\n",
" [ 1 3 36 3 8 128 0 22 1 4]\n",
" [ 17 4 1 6 8 1 134 0 0 0]\n",
" [ 1 1 23 1 0 14 1 141 0 0]\n",
" [ 4 16 1 1 7 0 3 0 127 0]\n",
" [ 0 7 3 1 31 0 1 3 1 75]]\n",
"val Loss: 0.8030 Acc: 0.7380\n",
"Confusion Matrix:\n",
"[[106 6 1 21 1 0 19 1 0 2]\n",
" [ 10 116 0 2 22 1 7 4 3 2]\n",
" [ 2 0 71 0 0 33 0 15 0 2]\n",
" [ 23 7 1 87 6 0 4 0 0 1]\n",
" [ 6 14 1 10 186 2 3 1 1 9]\n",
" [ 1 0 22 1 1 85 0 11 0 2]\n",
" [ 9 1 0 6 2 0 83 0 0 0]\n",
" [ 0 0 6 0 1 10 0 92 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 3 1 2 7 0 1 0 0 54]]\n",
"\n",
"Epoch 96/119\n",
"----------\n",
"train Loss: 0.9538 Acc: 0.6565\n",
"Confusion Matrix:\n",
"[[165 15 2 26 24 1 15 0 9 6]\n",
" [ 11 162 1 9 66 3 5 4 10 10]\n",
" [ 2 3 125 0 7 46 3 18 1 1]\n",
" [ 37 4 0 130 24 0 16 2 0 3]\n",
" [ 6 45 4 18 284 7 4 2 4 18]\n",
" [ 0 4 45 1 13 118 0 24 0 1]\n",
" [ 23 6 0 9 4 1 126 0 1 1]\n",
" [ 0 1 26 0 1 19 0 134 0 1]\n",
" [ 1 8 2 6 20 0 2 2 117 1]\n",
" [ 2 7 0 5 22 1 1 1 1 82]]\n",
"val Loss: 0.8026 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[105 4 1 22 1 0 18 3 1 2]\n",
" [ 9 113 0 2 24 1 7 4 5 2]\n",
" [ 2 0 66 0 0 30 0 23 0 2]\n",
" [ 15 4 1 93 9 0 5 1 0 1]\n",
" [ 6 15 1 11 183 3 3 1 2 8]\n",
" [ 0 0 20 1 1 79 0 21 0 1]\n",
" [ 7 2 0 6 2 0 84 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 3 4 0 0 0 86 0]\n",
" [ 4 3 2 4 6 0 1 0 0 52]]\n",
"\n",
"Epoch 97/119\n",
"----------\n",
"train Loss: 0.9432 Acc: 0.6724\n",
"Confusion Matrix:\n",
"[[168 13 3 29 19 0 20 0 7 4]\n",
" [ 13 180 3 8 54 1 6 0 7 9]\n",
" [ 2 2 116 1 3 49 3 26 0 4]\n",
" [ 39 5 3 126 20 2 15 3 2 1]\n",
" [ 11 38 6 17 283 5 5 2 9 16]\n",
" [ 2 1 31 1 6 143 0 21 0 1]\n",
" [ 19 5 1 15 4 0 122 1 0 4]\n",
" [ 1 0 25 0 1 22 2 128 0 3]\n",
" [ 4 12 0 5 9 0 1 1 127 0]\n",
" [ 0 8 1 5 22 0 0 0 1 85]]\n",
"val Loss: 0.8058 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[114 5 1 18 2 0 13 1 1 2]\n",
" [ 8 112 0 2 28 1 7 4 3 2]\n",
" [ 2 0 67 0 1 35 0 16 0 2]\n",
" [ 26 4 1 84 9 0 4 0 0 1]\n",
" [ 6 13 0 10 186 2 4 1 1 10]\n",
" [ 0 0 26 0 1 81 0 13 0 2]\n",
" [ 10 1 0 6 2 0 82 0 0 0]\n",
" [ 0 0 5 0 1 10 0 93 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 1 1 2 8 0 1 0 0 55]]\n",
"\n",
"Epoch 98/119\n",
"----------\n",
"train Loss: 0.9507 Acc: 0.6674\n",
"Confusion Matrix:\n",
"[[166 9 2 37 17 0 22 2 3 5]\n",
" [ 10 185 1 12 55 2 3 3 3 7]\n",
" [ 3 2 123 0 7 47 2 18 0 4]\n",
" [ 40 8 2 125 20 0 17 2 1 1]\n",
" [ 13 49 4 16 273 8 3 2 8 16]\n",
" [ 2 4 34 1 9 131 1 24 0 0]\n",
" [ 24 8 1 12 6 0 119 1 0 0]\n",
" [ 0 0 23 0 0 21 2 136 0 0]\n",
" [ 2 16 1 2 15 0 2 0 120 1]\n",
" [ 1 8 0 2 18 0 1 0 3 89]]\n",
"val Loss: 0.8052 Acc: 0.7280\n",
"Confusion Matrix:\n",
"[[114 4 1 20 2 0 12 1 1 2]\n",
" [ 9 113 0 1 25 1 7 4 5 2]\n",
" [ 2 0 68 0 0 36 0 15 0 2]\n",
" [ 27 5 0 84 8 0 4 0 0 1]\n",
" [ 6 14 0 11 180 2 3 1 6 10]\n",
" [ 1 0 23 1 1 86 0 9 0 2]\n",
" [ 10 2 0 7 2 0 80 0 0 0]\n",
" [ 0 0 8 0 1 14 0 86 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 1 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 99/119\n",
"----------\n",
"train Loss: 0.9490 Acc: 0.6752\n",
"Confusion Matrix:\n",
"[[169 10 1 34 17 0 18 2 4 8]\n",
" [ 16 172 1 6 59 2 7 5 4 9]\n",
" [ 4 2 125 2 6 38 3 23 0 3]\n",
" [ 32 7 2 138 19 3 11 1 1 2]\n",
" [ 9 49 6 16 278 3 5 2 7 17]\n",
" [ 0 2 41 2 8 128 0 23 0 2]\n",
" [ 22 2 2 11 6 0 128 0 0 0]\n",
" [ 0 1 15 1 0 16 0 148 1 0]\n",
" [ 4 11 0 3 16 0 2 0 122 1]\n",
" [ 5 5 2 3 27 0 3 1 0 76]]\n",
"val Loss: 0.8103 Acc: 0.7471\n",
"Confusion Matrix:\n",
"[[109 5 1 22 2 1 12 2 1 2]\n",
" [ 5 113 0 2 28 1 7 4 5 2]\n",
" [ 1 0 68 0 1 35 0 16 0 2]\n",
" [ 15 3 1 95 10 0 3 1 0 1]\n",
" [ 5 10 1 10 189 3 3 1 2 9]\n",
" [ 0 0 22 0 1 88 0 10 0 2]\n",
" [ 7 2 1 8 4 0 79 0 0 0]\n",
" [ 0 0 4 0 1 10 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 8 0 1 0 0 57]]\n",
"\n",
"Epoch 100/119\n",
"----------\n",
"train Loss: 0.9502 Acc: 0.6738\n",
"Confusion Matrix:\n",
"[[170 17 3 35 14 0 18 0 3 3]\n",
" [ 17 174 1 6 60 2 5 4 7 5]\n",
" [ 2 3 122 2 9 40 2 22 0 4]\n",
" [ 40 5 1 134 20 1 13 1 0 1]\n",
" [ 12 36 8 23 273 5 8 1 9 17]\n",
" [ 1 2 41 2 9 123 0 25 0 3]\n",
" [ 18 3 0 12 3 0 133 2 0 0]\n",
" [ 0 0 21 0 0 16 2 143 0 0]\n",
" [ 7 13 0 1 18 1 0 0 119 0]\n",
" [ 3 3 1 3 18 1 1 2 0 90]]\n",
"val Loss: 0.8146 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[110 7 1 18 3 0 13 1 2 2]\n",
" [ 5 118 0 0 26 1 7 4 4 2]\n",
" [ 2 0 68 0 1 29 0 20 1 2]\n",
" [ 25 3 1 80 15 0 4 0 0 1]\n",
" [ 5 14 0 8 188 1 3 1 3 10]\n",
" [ 0 1 21 1 2 80 0 16 0 2]\n",
" [ 10 3 0 7 3 0 78 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 2 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 101/119\n",
"----------\n",
"train Loss: 0.9421 Acc: 0.6779\n",
"Confusion Matrix:\n",
"[[166 13 0 30 17 1 26 0 5 5]\n",
" [ 16 171 3 8 56 3 7 2 5 10]\n",
" [ 1 1 123 1 8 41 2 25 0 4]\n",
" [ 34 12 0 128 23 3 14 0 0 2]\n",
" [ 4 47 6 15 280 4 7 0 9 20]\n",
" [ 0 6 35 1 9 129 4 20 0 2]\n",
" [ 23 5 0 10 7 0 125 0 1 0]\n",
" [ 0 0 16 0 2 14 1 148 0 1]\n",
" [ 3 11 0 2 11 1 1 0 129 1]\n",
" [ 2 6 1 4 14 3 1 0 0 91]]\n",
"val Loss: 0.8072 Acc: 0.7265\n",
"Confusion Matrix:\n",
"[[117 6 1 15 2 0 11 1 2 2]\n",
" [ 8 115 0 0 26 1 7 3 5 2]\n",
" [ 2 0 65 0 1 35 0 17 1 2]\n",
" [ 29 5 0 81 9 0 4 0 0 1]\n",
" [ 7 15 0 10 180 1 3 1 6 10]\n",
" [ 2 1 21 1 2 86 0 8 0 2]\n",
" [ 13 4 0 7 2 0 75 0 0 0]\n",
" [ 0 0 5 0 1 13 0 90 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 1 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 102/119\n",
"----------\n",
"train Loss: 0.9328 Acc: 0.6811\n",
"Confusion Matrix:\n",
"[[177 16 1 32 16 1 11 1 3 5]\n",
" [ 12 178 2 2 57 2 5 3 8 12]\n",
" [ 4 3 117 0 7 51 0 20 0 4]\n",
" [ 31 5 0 133 23 1 19 2 1 1]\n",
" [ 8 40 3 18 282 6 7 1 11 16]\n",
" [ 3 0 25 3 13 133 1 24 1 3]\n",
" [ 17 4 1 16 2 1 128 0 0 2]\n",
" [ 0 2 19 0 0 16 0 144 0 1]\n",
" [ 6 16 1 6 6 1 0 1 120 2]\n",
" [ 3 7 2 2 19 2 0 0 2 85]]\n",
"val Loss: 0.8154 Acc: 0.7280\n",
"Confusion Matrix:\n",
"[[111 6 1 20 3 0 10 1 3 2]\n",
" [ 5 114 0 0 28 1 7 4 6 2]\n",
" [ 1 1 70 0 1 33 0 15 0 2]\n",
" [ 23 3 1 85 12 0 4 0 0 1]\n",
" [ 6 15 0 8 188 1 3 1 3 8]\n",
" [ 0 1 29 1 1 80 0 9 0 2]\n",
" [ 10 5 0 7 4 0 75 0 0 0]\n",
" [ 0 1 8 0 1 11 0 88 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 4 1 2 8 0 0 0 0 56]]\n",
"\n",
"Epoch 103/119\n",
"----------\n",
"train Loss: 0.9296 Acc: 0.6793\n",
"Confusion Matrix:\n",
"[[178 13 0 32 11 1 18 2 3 5]\n",
" [ 13 179 1 7 59 1 7 0 8 6]\n",
" [ 4 2 126 2 6 38 2 23 0 3]\n",
" [ 33 8 2 133 18 1 16 1 2 2]\n",
" [ 6 52 3 18 273 4 9 0 8 19]\n",
" [ 1 2 32 0 5 146 0 19 0 1]\n",
" [ 17 6 1 9 7 0 129 1 0 1]\n",
" [ 0 2 25 0 2 18 3 131 0 1]\n",
" [ 3 15 0 8 9 0 0 2 120 2]\n",
" [ 3 7 1 4 25 1 0 3 0 78]]\n",
"val Loss: 0.7997 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[112 3 1 19 1 0 15 3 1 2]\n",
" [ 11 113 0 1 25 1 7 4 3 2]\n",
" [ 2 0 71 0 0 24 0 24 0 2]\n",
" [ 23 5 1 89 7 0 4 0 0 0]\n",
" [ 6 14 1 10 182 3 3 1 3 10]\n",
" [ 0 0 29 1 1 73 0 17 0 2]\n",
" [ 10 1 1 6 2 0 81 0 0 0]\n",
" [ 0 0 3 0 1 9 0 96 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 3 1 2 8 0 1 0 0 54]]\n",
"\n",
"Epoch 104/119\n",
"----------\n",
"train Loss: 0.9591 Acc: 0.6642\n",
"Confusion Matrix:\n",
"[[164 14 3 32 14 1 25 3 5 2]\n",
" [ 14 172 1 5 55 2 5 2 12 13]\n",
" [ 2 1 129 0 4 43 4 20 0 3]\n",
" [ 37 7 1 130 17 2 15 1 3 3]\n",
" [ 10 49 7 18 272 1 7 3 8 17]\n",
" [ 1 2 41 1 11 121 1 25 1 2]\n",
" [ 15 6 1 16 6 1 125 0 0 1]\n",
" [ 0 0 17 0 2 14 1 146 0 2]\n",
" [ 4 12 1 1 14 1 0 0 125 1]\n",
" [ 2 5 2 3 27 1 3 1 2 76]]\n",
"val Loss: 0.8144 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[111 5 1 20 1 0 12 2 3 2]\n",
" [ 6 117 0 0 26 1 7 4 4 2]\n",
" [ 2 0 64 0 1 35 0 19 0 2]\n",
" [ 23 5 1 84 11 0 3 1 0 1]\n",
" [ 6 17 1 10 179 3 3 1 4 9]\n",
" [ 0 1 20 0 1 87 0 12 0 2]\n",
" [ 10 4 0 7 3 0 77 0 0 0]\n",
" [ 0 0 2 0 1 11 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 105/119\n",
"----------\n",
"train Loss: 0.9550 Acc: 0.6611\n",
"Confusion Matrix:\n",
"[[164 14 3 39 11 0 20 3 6 3]\n",
" [ 18 178 1 7 50 3 4 1 8 11]\n",
" [ 3 1 122 0 5 41 1 29 0 4]\n",
" [ 37 9 1 134 22 1 11 1 0 0]\n",
" [ 7 51 5 26 265 3 6 1 10 18]\n",
" [ 2 3 46 3 10 120 1 18 1 2]\n",
" [ 19 8 0 6 6 0 128 0 2 2]\n",
" [ 0 0 18 0 0 23 1 139 0 1]\n",
" [ 4 12 1 4 10 0 1 0 126 1]\n",
" [ 2 8 3 4 25 0 1 1 1 77]]\n",
"val Loss: 0.8078 Acc: 0.7372\n",
"Confusion Matrix:\n",
"[[110 7 1 19 3 0 12 1 2 2]\n",
" [ 6 113 0 0 31 1 7 3 4 2]\n",
" [ 2 0 66 0 1 34 0 18 0 2]\n",
" [ 22 3 0 86 13 0 4 0 0 1]\n",
" [ 5 12 0 8 187 2 3 1 2 13]\n",
" [ 0 0 21 1 1 86 0 11 0 3]\n",
" [ 10 3 0 7 3 0 78 0 0 0]\n",
" [ 0 0 4 0 1 10 0 93 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 6 0 0 0 0 60]]\n",
"\n",
"Epoch 106/119\n",
"----------\n",
"train Loss: 0.9527 Acc: 0.6706\n",
"Confusion Matrix:\n",
"[[181 9 3 27 15 0 15 4 4 5]\n",
" [ 10 186 4 8 50 3 3 3 4 10]\n",
" [ 2 1 117 2 9 54 1 16 0 4]\n",
" [ 48 6 2 124 20 3 8 0 1 4]\n",
" [ 14 47 5 22 273 4 7 2 5 13]\n",
" [ 2 4 38 1 8 129 1 22 0 1]\n",
" [ 16 7 0 15 8 0 125 0 0 0]\n",
" [ 0 1 18 1 2 16 1 142 1 0]\n",
" [ 5 14 0 5 10 1 1 0 122 1]\n",
" [ 0 5 3 1 34 1 1 2 0 75]]\n",
"val Loss: 0.8032 Acc: 0.7196\n",
"Confusion Matrix:\n",
"[[117 3 1 12 1 0 17 3 1 2]\n",
" [ 8 116 0 0 23 1 7 4 5 3]\n",
" [ 2 0 68 0 0 26 0 25 0 2]\n",
" [ 31 8 1 78 6 0 4 1 0 0]\n",
" [ 8 20 1 10 172 1 3 1 5 12]\n",
" [ 0 0 29 0 1 67 0 24 0 2]\n",
" [ 9 2 0 6 2 0 82 0 0 0]\n",
" [ 0 0 3 0 1 6 0 99 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 2 1 2 7 0 1 0 0 56]]\n",
"\n",
"Epoch 107/119\n",
"----------\n",
"train Loss: 0.9430 Acc: 0.6683\n",
"Confusion Matrix:\n",
"[[172 7 0 29 21 1 24 1 6 2]\n",
" [ 14 176 1 10 54 2 6 5 7 6]\n",
" [ 0 2 123 1 4 51 2 20 0 3]\n",
" [ 40 8 2 131 12 1 17 2 1 2]\n",
" [ 14 47 6 18 268 7 5 0 8 19]\n",
" [ 1 3 40 1 8 126 0 24 0 3]\n",
" [ 18 4 1 11 7 0 130 0 0 0]\n",
" [ 0 0 24 1 1 17 0 139 0 0]\n",
" [ 2 15 0 6 11 0 0 0 125 0]\n",
" [ 2 5 0 3 29 2 0 2 0 79]]\n",
"val Loss: 0.8067 Acc: 0.7296\n",
"Confusion Matrix:\n",
"[[110 5 1 19 2 0 15 3 0 2]\n",
" [ 8 116 0 1 25 1 7 4 3 2]\n",
" [ 2 0 67 0 0 29 0 23 0 2]\n",
" [ 24 5 1 85 8 0 4 1 0 1]\n",
" [ 5 19 1 10 177 3 4 1 3 10]\n",
" [ 0 0 19 0 1 82 0 19 0 2]\n",
" [ 9 3 0 7 2 0 80 0 0 0]\n",
" [ 0 0 2 0 1 8 0 98 0 0]\n",
" [ 0 2 0 2 6 0 0 0 85 0]\n",
" [ 3 3 1 2 7 0 1 0 0 55]]\n",
"\n",
"Epoch 108/119\n",
"----------\n",
"train Loss: 0.9633 Acc: 0.6670\n",
"Confusion Matrix:\n",
"[[172 14 1 32 17 1 15 1 4 6]\n",
" [ 10 176 4 6 61 0 4 5 7 8]\n",
" [ 1 0 120 1 9 47 3 21 0 4]\n",
" [ 46 9 2 116 25 1 13 1 3 0]\n",
" [ 14 45 5 19 267 6 7 1 9 19]\n",
" [ 1 4 36 2 8 135 1 19 0 0]\n",
" [ 21 6 2 8 6 0 127 0 0 1]\n",
" [ 0 0 17 0 1 18 2 142 0 2]\n",
" [ 5 12 0 4 5 0 0 1 131 1]\n",
" [ 2 4 1 4 25 0 1 3 2 80]]\n",
"val Loss: 0.7972 Acc: 0.7334\n",
"Confusion Matrix:\n",
"[[115 5 1 17 2 0 13 1 1 2]\n",
" [ 9 118 0 1 23 1 7 3 3 2]\n",
" [ 2 0 71 0 0 33 0 15 0 2]\n",
" [ 27 6 1 83 7 0 4 0 0 1]\n",
" [ 6 17 0 10 183 1 3 1 2 10]\n",
" [ 2 1 26 1 1 77 0 13 0 2]\n",
" [ 10 2 0 6 2 0 81 0 0 0]\n",
" [ 0 0 6 0 1 8 0 94 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 4 1 2 8 0 1 0 0 52]]\n",
"\n",
"Epoch 109/119\n",
"----------\n",
"train Loss: 0.9473 Acc: 0.6747\n",
"Confusion Matrix:\n",
"[[170 12 5 32 14 2 21 1 4 2]\n",
" [ 17 179 1 5 47 3 2 5 13 9]\n",
" [ 1 1 130 0 4 42 3 21 0 4]\n",
" [ 40 2 2 124 26 2 15 1 0 4]\n",
" [ 11 47 6 22 271 5 5 1 5 19]\n",
" [ 0 2 46 2 3 127 1 24 0 1]\n",
" [ 14 3 1 10 5 2 134 0 0 2]\n",
" [ 0 0 15 1 2 18 0 144 0 2]\n",
" [ 5 8 1 2 16 0 1 0 124 2]\n",
" [ 1 9 2 4 24 2 0 0 0 80]]\n",
"val Loss: 0.8060 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[111 5 1 19 3 0 13 1 2 2]\n",
" [ 6 111 0 0 31 1 7 4 5 2]\n",
" [ 2 0 69 0 0 31 0 19 0 2]\n",
" [ 23 3 1 87 10 0 4 0 0 1]\n",
" [ 6 13 0 10 185 2 3 1 3 10]\n",
" [ 0 1 27 1 1 76 0 15 0 2]\n",
" [ 9 3 0 7 3 0 79 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 1 0 0 58]]\n",
"\n",
"Epoch 110/119\n",
"----------\n",
"train Loss: 0.9581 Acc: 0.6661\n",
"Confusion Matrix:\n",
"[[171 15 1 33 18 0 16 1 6 2]\n",
" [ 14 194 1 4 50 1 4 0 4 9]\n",
" [ 1 4 117 0 4 53 3 21 0 3]\n",
" [ 39 4 1 124 25 1 15 1 4 2]\n",
" [ 12 51 8 17 281 1 7 1 3 11]\n",
" [ 3 4 37 2 6 126 0 28 0 0]\n",
" [ 27 4 1 14 4 0 120 0 1 0]\n",
" [ 0 1 20 0 0 21 2 137 0 1]\n",
" [ 2 17 1 3 14 0 1 1 119 1]\n",
" [ 3 10 3 2 24 3 0 1 1 75]]\n",
"val Loss: 0.8027 Acc: 0.7372\n",
"Confusion Matrix:\n",
"[[106 6 1 22 1 0 18 1 0 2]\n",
" [ 8 118 0 2 22 1 7 4 3 2]\n",
" [ 2 0 64 0 0 33 0 22 0 2]\n",
" [ 14 7 1 96 6 0 4 0 0 1]\n",
" [ 6 18 1 10 180 2 3 1 2 10]\n",
" [ 0 0 20 1 1 82 0 17 0 2]\n",
" [ 6 2 0 6 2 0 85 0 0 0]\n",
" [ 0 0 3 0 1 8 0 97 0 0]\n",
" [ 0 4 0 3 3 0 0 0 85 0]\n",
" [ 4 4 1 3 7 0 1 0 0 52]]\n",
"\n",
"Epoch 111/119\n",
"----------\n",
"train Loss: 0.9446 Acc: 0.6679\n",
"Confusion Matrix:\n",
"[[172 8 4 34 16 2 16 0 5 6]\n",
" [ 12 175 0 9 56 7 6 4 4 8]\n",
" [ 2 3 116 1 6 47 3 24 0 4]\n",
" [ 30 5 2 137 21 0 15 0 3 3]\n",
" [ 10 50 6 12 275 2 9 1 9 18]\n",
" [ 1 5 43 0 8 121 0 27 0 1]\n",
" [ 19 4 0 11 9 2 124 1 0 1]\n",
" [ 0 1 25 0 0 15 1 139 0 1]\n",
" [ 2 9 0 1 18 0 1 0 128 0]\n",
" [ 2 4 2 3 26 1 3 0 0 81]]\n",
"val Loss: 0.8115 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[113 5 1 20 2 0 11 3 0 2]\n",
" [ 7 114 0 0 25 2 7 5 5 2]\n",
" [ 1 0 65 0 0 34 0 21 0 2]\n",
" [ 20 8 1 88 7 0 3 1 0 1]\n",
" [ 6 13 1 11 184 3 2 1 2 10]\n",
" [ 0 0 19 0 1 86 0 15 0 2]\n",
" [ 11 4 1 7 3 0 75 0 0 0]\n",
" [ 0 0 3 0 1 9 0 96 0 0]\n",
" [ 0 2 0 2 6 0 0 0 85 0]\n",
" [ 2 4 2 2 8 0 1 0 0 53]]\n",
"\n",
"Epoch 112/119\n",
"----------\n",
"train Loss: 0.9606 Acc: 0.6729\n",
"Confusion Matrix:\n",
"[[185 8 3 27 17 0 17 0 2 4]\n",
" [ 17 176 0 9 53 3 8 2 5 8]\n",
" [ 2 2 118 2 9 49 2 20 0 2]\n",
" [ 33 5 2 131 25 1 15 1 3 0]\n",
" [ 12 49 9 15 276 3 6 0 3 19]\n",
" [ 2 2 46 0 7 131 0 17 0 1]\n",
" [ 15 8 3 11 9 0 124 0 0 1]\n",
" [ 0 1 17 1 1 20 0 140 0 2]\n",
" [ 3 11 0 3 16 0 1 0 124 1]\n",
" [ 0 13 1 2 26 2 1 2 1 74]]\n",
"val Loss: 0.8120 Acc: 0.7349\n",
"Confusion Matrix:\n",
"[[110 8 1 19 3 0 12 1 1 2]\n",
" [ 5 121 0 0 23 1 7 3 4 3]\n",
" [ 2 0 67 0 1 33 0 18 0 2]\n",
" [ 25 6 0 84 9 0 4 0 0 1]\n",
" [ 5 17 0 8 185 1 3 1 3 10]\n",
" [ 0 1 21 1 1 85 0 11 0 3]\n",
" [ 11 5 0 7 4 0 74 0 0 0]\n",
" [ 0 1 5 0 1 10 0 91 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 2 1 1 2 7 0 0 0 0 59]]\n",
"\n",
"Epoch 113/119\n",
"----------\n",
"train Loss: 0.9645 Acc: 0.6733\n",
"Confusion Matrix:\n",
"[[171 12 2 31 18 1 19 1 5 3]\n",
" [ 11 187 3 7 50 1 3 3 8 8]\n",
" [ 0 4 136 0 6 42 2 14 0 2]\n",
" [ 52 5 1 111 24 3 16 1 2 1]\n",
" [ 13 37 5 21 280 4 8 1 6 17]\n",
" [ 1 4 32 3 6 133 1 21 0 5]\n",
" [ 14 5 0 15 10 0 124 2 0 1]\n",
" [ 0 0 13 2 1 20 1 143 0 2]\n",
" [ 3 16 0 2 15 0 1 0 120 2]\n",
" [ 2 6 2 4 25 0 5 2 1 75]]\n",
"val Loss: 0.8083 Acc: 0.7311\n",
"Confusion Matrix:\n",
"[[114 5 1 17 2 0 14 1 1 2]\n",
" [ 8 111 0 0 28 0 8 4 6 2]\n",
" [ 2 0 69 0 1 31 0 18 0 2]\n",
" [ 22 4 0 89 9 0 4 0 0 1]\n",
" [ 6 13 0 10 185 1 3 1 3 11]\n",
" [ 2 0 30 1 1 71 0 15 0 3]\n",
" [ 11 1 0 6 3 0 80 0 0 0]\n",
" [ 0 0 4 0 1 8 0 95 0 1]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 3 1 1 2 8 0 1 0 0 56]]\n",
"\n",
"Epoch 114/119\n",
"----------\n",
"train Loss: 0.9313 Acc: 0.6843\n",
"Confusion Matrix:\n",
"[[161 16 1 36 11 2 23 1 6 6]\n",
" [ 13 189 0 9 47 1 7 5 5 5]\n",
" [ 2 1 134 2 3 42 3 15 1 3]\n",
" [ 34 4 1 134 20 2 16 3 1 1]\n",
" [ 12 39 4 17 279 7 3 3 13 15]\n",
" [ 0 3 39 1 11 133 3 15 0 1]\n",
" [ 19 5 1 10 8 0 127 1 0 0]\n",
" [ 0 3 16 1 1 17 0 144 0 0]\n",
" [ 4 12 1 3 13 0 0 0 125 1]\n",
" [ 1 2 3 4 26 0 4 2 2 78]]\n",
"val Loss: 0.8091 Acc: 0.7326\n",
"Confusion Matrix:\n",
"[[107 6 1 19 3 0 15 2 2 2]\n",
" [ 5 118 0 0 25 1 7 4 5 2]\n",
" [ 2 0 68 0 0 32 0 19 0 2]\n",
" [ 17 6 1 91 8 0 4 1 0 1]\n",
" [ 5 17 1 10 178 1 3 1 6 11]\n",
" [ 0 0 26 0 1 76 0 18 0 2]\n",
" [ 8 4 0 7 3 0 79 0 0 0]\n",
" [ 0 0 4 0 1 7 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 1 1 2 7 0 1 0 0 58]]\n",
"\n",
"Epoch 115/119\n",
"----------\n",
"train Loss: 0.9235 Acc: 0.6820\n",
"Confusion Matrix:\n",
"[[179 11 2 29 15 1 15 1 4 6]\n",
" [ 17 171 0 12 53 5 5 1 5 12]\n",
" [ 1 4 134 0 6 39 2 18 0 2]\n",
" [ 35 4 1 132 19 1 20 0 2 2]\n",
" [ 10 51 4 18 285 6 1 0 3 14]\n",
" [ 1 2 47 2 11 118 2 19 0 4]\n",
" [ 16 4 0 12 6 1 131 0 0 1]\n",
" [ 0 1 18 0 1 16 1 145 0 0]\n",
" [ 2 18 0 3 10 0 1 2 122 1]\n",
" [ 2 3 2 2 21 3 3 3 1 82]]\n",
"val Loss: 0.8094 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[112 6 1 18 2 0 13 1 2 2]\n",
" [ 10 113 0 0 26 1 7 2 5 3]\n",
" [ 1 1 71 0 0 35 0 13 0 2]\n",
" [ 25 6 0 86 7 0 4 0 0 1]\n",
" [ 7 14 0 10 184 1 3 1 3 10]\n",
" [ 0 1 24 0 1 87 0 8 0 2]\n",
" [ 11 2 0 6 3 0 79 0 0 0]\n",
" [ 0 1 8 0 1 17 0 81 0 1]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 3 2 1 2 7 0 0 0 0 57]]\n",
"\n",
"Epoch 116/119\n",
"----------\n",
"train Loss: 0.9397 Acc: 0.6833\n",
"Confusion Matrix:\n",
"[[168 18 2 35 15 0 16 2 5 2]\n",
" [ 11 181 3 6 49 5 6 4 5 11]\n",
" [ 5 3 117 1 5 48 1 25 0 1]\n",
" [ 41 6 1 135 17 1 12 1 1 1]\n",
" [ 11 41 6 19 285 3 6 0 8 13]\n",
" [ 2 0 38 1 8 131 0 21 0 5]\n",
" [ 20 4 1 11 4 0 130 0 0 1]\n",
" [ 0 0 10 2 4 17 1 148 0 0]\n",
" [ 5 18 0 3 9 0 0 0 123 1]\n",
" [ 2 4 1 3 23 0 2 3 0 84]]\n",
"val Loss: 0.8179 Acc: 0.7341\n",
"Confusion Matrix:\n",
"[[112 7 1 17 3 0 11 1 3 2]\n",
" [ 5 120 0 0 22 1 7 4 5 3]\n",
" [ 1 1 63 0 0 34 0 22 0 2]\n",
" [ 27 6 0 83 8 0 3 1 0 1]\n",
" [ 5 18 0 10 178 1 3 1 6 11]\n",
" [ 0 0 17 1 1 86 0 16 0 2]\n",
" [ 11 4 0 7 2 0 77 0 0 0]\n",
" [ 0 0 2 0 1 9 0 97 0 0]\n",
" [ 0 2 0 2 4 0 0 0 87 0]\n",
" [ 2 2 1 2 6 0 1 0 0 58]]\n",
"\n",
"Epoch 117/119\n",
"----------\n",
"train Loss: 0.9435 Acc: 0.6797\n",
"Confusion Matrix:\n",
"[[172 9 4 24 23 2 22 0 7 0]\n",
" [ 16 182 2 8 50 2 7 2 3 9]\n",
" [ 4 0 125 2 5 47 0 19 0 4]\n",
" [ 37 7 2 127 20 2 12 1 5 3]\n",
" [ 3 44 7 17 278 3 10 1 11 18]\n",
" [ 0 3 34 1 3 137 2 23 0 3]\n",
" [ 22 4 1 13 5 0 125 0 0 1]\n",
" [ 0 2 12 0 1 20 0 146 0 1]\n",
" [ 3 15 3 2 11 0 0 0 123 2]\n",
" [ 2 7 1 2 26 0 3 1 1 79]]\n",
"val Loss: 0.8101 Acc: 0.7349\n",
"Confusion Matrix:\n",
"[[108 5 1 20 3 0 14 3 1 2]\n",
" [ 4 121 0 0 22 2 7 4 5 2]\n",
" [ 0 1 69 0 0 29 0 22 0 2]\n",
" [ 22 7 1 83 11 0 3 1 0 1]\n",
" [ 3 19 1 8 182 3 3 1 5 8]\n",
" [ 0 0 22 0 1 78 0 20 0 2]\n",
" [ 8 4 0 7 3 0 79 0 0 0]\n",
" [ 0 0 2 0 1 7 0 99 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 1 3 1 2 7 0 1 0 0 57]]\n",
"\n",
"Epoch 118/119\n",
"----------\n",
"train Loss: 0.9409 Acc: 0.6783\n",
"Confusion Matrix:\n",
"[[173 22 0 29 14 0 14 2 3 6]\n",
" [ 11 183 2 7 56 1 5 3 4 9]\n",
" [ 2 0 122 4 3 50 3 19 0 3]\n",
" [ 37 7 2 123 29 1 15 0 2 0]\n",
" [ 8 44 3 22 288 3 6 2 3 13]\n",
" [ 0 2 39 2 10 126 0 25 0 2]\n",
" [ 23 4 3 14 5 0 120 0 1 1]\n",
" [ 0 0 22 0 1 10 1 148 0 0]\n",
" [ 4 17 2 2 13 0 0 1 118 2]\n",
" [ 4 6 3 2 15 0 0 2 0 90]]\n",
"val Loss: 0.7982 Acc: 0.7303\n",
"Confusion Matrix:\n",
"[[112 4 1 19 2 0 14 2 1 2]\n",
" [ 9 110 0 0 28 1 7 4 6 2]\n",
" [ 2 0 72 0 0 31 0 16 0 2]\n",
" [ 20 3 1 91 9 0 4 0 0 1]\n",
" [ 6 13 1 10 184 2 3 1 3 10]\n",
" [ 0 0 34 0 1 73 0 13 0 2]\n",
" [ 11 1 0 5 2 0 82 0 0 0]\n",
" [ 0 0 7 0 1 10 0 91 0 0]\n",
" [ 0 2 0 2 5 0 0 0 86 0]\n",
" [ 4 2 1 2 7 0 1 0 0 55]]\n",
"\n",
"Epoch 119/119\n",
"----------\n",
"train Loss: 0.9277 Acc: 0.6752\n",
"Confusion Matrix:\n",
"[[174 9 3 33 16 0 20 1 5 2]\n",
" [ 13 180 0 6 52 4 7 2 7 10]\n",
" [ 3 3 121 2 5 45 1 24 0 2]\n",
" [ 36 4 1 138 14 1 17 2 1 2]\n",
" [ 9 44 4 22 277 6 9 2 5 14]\n",
" [ 0 3 35 0 6 134 0 26 0 2]\n",
" [ 21 8 0 11 7 0 120 3 1 0]\n",
" [ 0 2 16 0 2 13 0 149 0 0]\n",
" [ 3 14 0 3 19 0 0 0 118 2]\n",
" [ 1 5 3 4 30 1 1 2 2 73]]\n",
"val Loss: 0.8012 Acc: 0.7433\n",
"Confusion Matrix:\n",
"[[108 5 1 21 1 0 17 1 1 2]\n",
" [ 9 115 0 1 24 1 7 4 4 2]\n",
" [ 2 0 68 0 0 33 0 18 0 2]\n",
" [ 15 5 1 95 8 0 4 0 0 1]\n",
" [ 6 16 1 10 181 2 3 1 3 10]\n",
" [ 0 0 20 0 1 87 0 13 0 2]\n",
" [ 7 2 0 7 2 0 83 0 0 0]\n",
" [ 0 0 4 0 1 9 0 95 0 0]\n",
" [ 0 3 0 2 4 0 0 0 86 0]\n",
" [ 3 3 1 2 7 0 1 0 0 55]]\n",
"\n",
"Training complete in 54m 31s\n",
"Best val Acc: 0.747135\n"
]
}
],
"source": [
"model_conv = train_model(model_conv, criterion, optimizer_conv,\n",
" exp_lr_scheduler, num_epochs=120)"
]
},
{
"cell_type": "code",
"source": [
"seconds = time.time()\n",
"print(\"Time in seconds since beginning of run:\", seconds)\n",
"local_time = time.ctime(seconds)\n",
"print(local_time)"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"id": "FrvMovYIufeS",
"outputId": "a23c921b-a90a-41a6-c35b-8c86fe1d85cd"
},
"execution_count": 24,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Time in seconds since beginning of run: 1685213497.6850653\n",
"Sat May 27 18:51:37 2023\n"
]
}
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {
"id": "qPq7-3EfW2lF",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 923
},
"outputId": "c4348cc1-9b1f-4fd4-a6bc-8008851863f3"
},
"outputs": [
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "iVBORw0KGgoAAAANSUhEUgAAASkAAACXCAYAAABJLq0zAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABCx0lEQVR4nO19eZRcZZn+c2u9te9L73uS7iwQAwGFEJQoybDIYIAMRwUHxZFFkTkeD3KG5XfEcQZ1HEQ5AyOoCOM4oA44ChqJQ4gsEUJi9vS+VVdXde379v3+6Hm/3OruLISQ7mq+55w+6bp1l+9W133yLs/7vhJjjEFAQEBggUI13wsQEBAQOB4ESQkICCxoCJISEBBY0BAkJSAgsKAhSEpAQGBBQ5CUgIDAgoYgKQEBgQUNQVICAgILGoKkBAQEFjRqjqT++Mc/QpIk/PGPf+TbbrzxRrS2ts7bmmZirjXOFwYHByFJEr71rW/N91IEahTz/X2uOZI6nfjGN76BX/3qV/O9jDnx/PPPY/369fB6vTAajWhvb8e1116LF154Yb6XJvAOcOONN0KSpBP+3HjjjQCAN954A7fccgvWrFkDrVYLSZKOe/5EIoH7778fZ511FsxmMwwGA1asWIGvfvWrGB8fP+6xP/rRj6rWIMsylixZgttuuw3BYPB0fQTvGpr5XsDpwGOPPYZKpfKOj/vGN76BzZs346qrrjr9i3oX+Na3voWvfOUrWL9+Pe666y4YjUb09vZi69at+NnPfoaNGzfO9xIFThKf//znsWHDBv56YGAA99xzD26++WasW7eOb+/o6AAA/OY3v8G///u/Y9WqVWhvb8fhw4ePee7+/n5s2LABw8PDuOaaa3DzzTdDp9Nhz549+OEPf4hf/vKXxz2e8P/+3/9DW1sbcrkcXnnlFTzyyCP4zW9+g71798JoNL6Luz9NYGcI5XKZZbPZd32ebdu2MQBs27Zt7/pcJpOJ3XDDDe/6PDPxbtZYLBaZ1WplH/3oR+d8PxgMvqPzDQwMMADswQcffMdrETj92LlzJwPAnnjiiTnfn5iYYJlMhjHG2K233sqO9YgWi0V21llnMaPRyLZv3z7r/Xg8zr72ta8ddy1PPPEEA8B27txZtf3OO+9kANjTTz/NGDu9z9yp4B25e/fddx8kScLBgwdx7bXXwmq1wuVy4Utf+hJyuVzVvpIk4bbbbsNTTz2F5cuXQ6/Xc1dlbGwMf/u3fwufzwe9Xo/ly5fj8ccfn3W90dFRXHXVVTCZTPB6vfjyl7+MfD4/a7+5YlKVSgX/+q//ipUrV0KWZXg8HmzcuBF//vOf+frS6TR+/OMfzzK534s1ZjIZHDx4EOFw+LifcTgcRiKRwAUXXDDn+16vt+p1LpfDfffdhyVLlkCWZdTV1eHqq69GX1/frGMfffRRdHR0QK/X49xzz8XOnTv5e8899xwkScKePXv4tmeffRaSJOHqq6+uOk93dzeuu+46/vqJJ57ARz7yEXi9Xuj1evT09OCRRx6Zdf3W1lZcfvnleOWVV7B27VrIsoz29nb85Cc/qdqP3JAdO3bgzjvvhMfjgclkwl//9V8jFArNOu8PfvAD/h2rr6/HrbfeilgsVrXPxRdfjBUrVmDPnj1Yv349jEYjOjs78cwzzwAA/vd//xfnnXceDAYDli5diq1bt1YdPzQ0hFtuuQVLly6FwWCAy+XCNddcg8HBwVnreTfw+XwwGAwn3O/ZZ5/F7t27cffdd+PCCy+c9b7VasUDDzxwSmv4yEc+AmDa6jsWWltbq54XwsUXX4yLL764atv3vvc9LF++HEajEQ6HA+eccw6efvrpk17PKbl71157LVpbW/GP//iPeO211/DQQw8hGo3O+rK99NJL+PnPf47bbrsNbrcbra2tCAaDOP/88zmJeTwe/Pa3v8VNN92ERCKBO+64AwCQzWZxySWXYHh4GF/84hdRX1+PJ598Ei+99NJJrfGmm27Cj370I2zatAmf/exnUSqVsH37drz22ms455xz8OSTT+Kzn/0s1q5di5tvvhnAUZP7vVjjG2+8gQ9/+MO49957cd999x1z3V6vFwaDAc8//zxuv/12OJ3OY+5bLpdx+eWX4w9/+AO2bNmCL33pS0gmk/j973+PvXv38vsBgKeffhrJZBKf//znIUkS/vmf/xlXX301+vv7odVqceGFF0KSJLz88stYtWoVAGD79u1QqVR45ZVX+HlCoRAOHjyI2267jW975JFHsHz5clx55ZXQaDR4/vnnccstt6BSqeDWW2+tWnNvby82b96Mm266CTfccAMef/xx3HjjjVizZg2WL19ete/tt98Oh8OBe++9F4ODg/jud7+L2267Df/5n//J97nvvvtw//33Y8OGDfjCF76AQ4cO4ZFHHsHOnTuxY8cOaLVavm80GsXll1+OLVu24JprrsEjjzyCLVu24KmnnsIdd9yBv/u7v8P111+PBx98EJs3b8bIyAgsFgsAYOfOnfjTn/6ELVu2oLGxEYODg3jkkUdw8cUXY//+/WfcLXruuecAAJ/61KdO+7npPziXy/Wuz/XYY4/hi1/8IjZv3syNmT179uD111/H9ddff3IneSdm17333ssAsCuvvLJq+y233MIAsN27d/NtAJhKpWL79u2r2vemm25idXV1LBwOV23fsmULs9ls3NT97ne/ywCwn//853yfdDrNOjs7Z5meN9xwA2tpaeGvX3rpJQaAffGLX5x1D5VKhf9+LHfvvVgjmcz33nvvrOvNxD333MMAMJPJxDZt2sQeeOAB9uabb87a7/HHH2cA2He+851j3ie5ey6Xi0UiEf7+f//3fzMA7Pnnn+fbli9fzq699lr++gMf+AC75pprGAB24MABxhhjv/jFL2b9renzUOLSSy9l7e3tVdtaWloYAPbyyy/zbZOTk0yv17O///u/59vIDdmwYUPV3+vLX/4yU6vVLBaL8WN1Oh372Mc+xsrlMt/v4YcfZgDY448/zretX7++yoVhjLGDBw/y7+lrr73Gt7/44ouzXLK57vHVV19lANhPfvKTWe8dCydy95Q4nru3evVqZrPZTvq6c4E+561bt7JQKMRGRkbYz372M+ZyuZjBYGCjo6OMsbndvZaWljmfnfXr17P169fz1x//+MfZ8uXL39U6Tym7N/N/x9tvvx3AdNBPifXr16Onp0dJiHj22WdxxRVXgDGGcDjMfy699FLE43G89dZb/Fx1dXXYvHkzP95oNHKr53ggN+Xee++d9d6JsiXv1RovvvhiMMaOa0UR7r//fjz99NNYvXo1XnzxRdx9991Ys2YNPvCBD+DAgQNV9+l2u/nnf7z7vO666+BwOPhrCtr29/dXbdu+fTsAIJlMYvfu3bj55pvhdrv59u3bt8Nut2PFihX8OKV7Eo/HEQ6HsX79evT39yMej1eto6enpypg7PF4sHTp0qp1EG6++eaq+1i3bh3K5TKGhoYAAFu3bkWhUMAdd9wBleroV/lzn/scrFYr/ud//qfqfGazGVu2bOGvly5dCrvdju7ubpx33nl8O/2uXJPyHovFIqamptDZ2Qm73c6/D2cSiUSCW3nvFhs2bIDH40FTUxO2bNkCs9mMX/7yl2hoaHjX57bb7RgdHa0KLbxTnBJJdXV1Vb3u6OiASqWa5Z+3tbVVvQ6FQojFYnj00Ufh8Xiqfj7zmc8AACYnJwFMxwA6OztnPWxLly494fr6+vpQX19/XFfpWDhTazwR/uZv/gbbt29HNBrF7373O1x//fXYtWsXrrjiCh7/6+vrw9KlS6HRnNhrb25urnpNhBWNRvm2devWIRAIoLe3F3/6058gSRI++MEPVpHX9u3bccEFF1SRwo4dO7BhwwaYTCbY7XZ4PB587WtfA4BZJDVzHbQW5TpOds1EVjM/b51Oh/b2dv4+obGxcdbfymazoampada2mZ9NNpvFPffcg6amJuj1erjdbng8HsRisVn3eCZgtVqRTCZPy7m+//3v4/e//z22bduG/fv3o7+/H5deeulpOfdXv/pVmM1mrF27Fl1dXbj11luxY8eOd3SO0yJBOJZ1MjMASDKBT37yk7jhhhvmPIbiIfOFhbZGq9WKj370o/joRz8KrVaLH//4x3j99dexfv36d3QetVo953am6B5NAdiXX34Z/f39+MAHPgCTyYR169bhoYceQiqVwq5du6oCsn19fbjkkkuwbNkyfOc730FTUxN0Oh1+85vf4F/+5V9mSUNOZh2nsu/J4FjnO5nr3H777XjiiSdwxx134IMf/CBsNhskScKWLVtOSf7ybrFs2TLs2rULIyMjs0j2nWLt2rU455xz3tExx3rmy+Vy1efZ3d2NQ4cO4de//jVeeOEFPPvss/jBD36Ae+65B/fff/9JXeuUSOrIkSNVVlJvby8qlcoJVd8ejwcWiwXlcrlKOzIXWlpasHfvXjDGqj6QQ4cOnXB9HR0dePHFFxGJRI5rTc31QZ+pNZ4KzjnnHPz4xz9GIBAAMH2fr7/+OorFYlWA+FTR3NyM5uZmbN++Hf39/dwtu+iii3DnnXfiv/7rv1Aul3HRRRfxY55//nnk83k899xzVZbPtm3b3vV6ToSWlhYA0593e3s7314oFDAwMHDCv987wTPPPIMbbrgB3/72t/m2XC43K4t4pnDFFVfgP/7jP/DTn/4Ud9111xm/vsPhmPPeh4aGqv4WAGAymXDdddfhuuuuQ6FQwNVXX40HHngAd911F2RZPuG1Tsnd+/73v1/1+nvf+x4AYNOmTcc9Tq1W4xOf+ASeffZZ7N27d9b7yvTyX/3VX2F8fJyniIHpNP6jjz56wvV94hOfAGNsTqZW/u9oMplmfdDv1RpPVoKQyWTw6quvzvneb3/7WwBH3ZtPfOITCIfDePjhh2fte6rWxrp16/DSSy/hjTfe4CR19tlnw2Kx4Jvf/CYMBgPWrFnD96f/NZXXi8fjeOKJJ07p+u8EGzZsgE6nw0MPPVR1/R/+8IeIx+O47LLLTtu11Gr1rM/0e9/7Hsrl8mm7xjvB5s2bsXLlSjzwwANzfl+SySTuvvvu9+z6HR0deO2111AoFPi2X//61xgZGanab2pqquq1TqdDT08PGGMoFosnda1TsqQGBgZw5ZVXYuPGjXj11Vfx05/+FNdffz3OOuusEx77zW9+E9u2bcN5552Hz33uc+jp6UEkEsFbb72FrVu3IhKJAJgOfj788MP49Kc/jTfffBN1dXV48sknTyrV++EPfxif+tSn8NBDD+HIkSPYuHEjKpUKtm/fjg9/+MM8fb5mzRps3boV3/nOd1BfX4+2tjacd95578kaT1aCkMlk8KEPfQjnn38+Nm7ciKamJsRiMfzqV7/C9u3bcdVVV2H16tUAgE9/+tP4yU9+gjvvvJOTSjqdxtatW3HLLbfg4x//+Ak/q5lYt24dnnrqKUiSxN0/tVqND33oQ3jxxRdx8cUXQ6fT8f0/9rGPQafT4YorrsDnP/95pFIpPPbYY/B6vdzie6/g8Xhw11134f7778fGjRtx5ZVX4tChQ/jBD36Ac889F5/85CdP27Uuv/xyPPnkk7DZbOjp6cGrr76KrVu3npY0vRJDQ0N48sknAYBr+r7+9a8DmLYcSXKg1Wrxi1/8Ahs2bMBFF12Ea6+9FhdccAG0Wi327duHp59+Gg6H45S1UifCZz/7WTzzzDPYuHEjrr32WvT19eGnP/1plewFmP5++P1+XHDBBfD5fDhw4AAefvhhXHbZZScf+H8nqUCSIOzfv59t3ryZWSwW5nA42G233TZLTQ6A3XrrrXOeJxgMsltvvZU1NTUxrVbL/H4/u+SSS9ijjz5atd/Q0BC78sormdFoZG63m33pS19iL7zwwgklCIwxViqV2IMPPsiWLVvGdDod83g8bNOmTVWp/IMHD7KLLrqIGQwGBqAqpXq613iyEoRiscgee+wxdtVVV7GWlham1+uZ0Whkq1evZg8++CDL5/NV+2cyGXb33XeztrY2vs7Nmzezvr4+xtjxFedzrWffvn0MAOvu7q7a/vWvf50BYP/wD/8w6zzPPfccW7VqFZNlmbW2trJ/+qd/4vKIgYEBvl9LSwu77LLLZh0/M219LCX0sZTPDz/8MFu2bBnTarXM5/OxL3zhCywajc66xlyp8GOtaeb3NxqNss985jPM7XYzs9nMLr30Unbw4MFjpuKPhRNJEOge5/pRfkbKdd1zzz1s5cqVzGg0MlmW2YoVK9hdd93FAoHAcddyrM/5WGua+bl/+9vfZg0NDUyv17MLLriA/fnPf571t/y3f/s3dtFFFzGXy8X0ej3r6OhgX/nKV1g8Hj/uNZWQGDt5v4CEc6FQCG63+2QPExAQEDhlvK+7IAgICCx8CJISEBBY0BAkJSAgsKDxjmJSAgICAmcawpISEBBY0BAkJSAgsKAhSEpAQGBBY1H0OK9VnKhtjMDCgwjhnnkIS0pAQGBBQ5CUgIDAgoYgKQEBgQUNQVICAgILGiJwLnDaoEwEiACzwOmCICmBd4VjZShpuyArgXcL4e4JnDKOJ6EQ8gqB0wVhSQmcMtj/9XaXJIlbTEROM0lKWFQCpwpBUgKnDCIo+l25XZCSwOmCICmBd4y5yElpVRHYjCk6grgETgWCpARmuWYnIhZJkqBSqSBJEtRqNVQqFQwGAwwGAxhjSCQSKBQKKJVKVcfQfLqZ5zyWFSZITQAQ/aTmFQsluEzTiOeyhOhfxhjfjwhKo9FAq9XC6XTC5/NBkiSUSiVUKhWMjo4iGo1yYiKSqlQqs9xBuiadnzFWNXBzIX1FF9Ja3i8QltT7HEQMZBHNJI9KpVIVFFe+r1Kp4HA40NbWhlAoBJ1OB41GA7VajebmZuTzeaTTaX6skoSU7iK9R++Xy2V+LWFVCQiSep9CSQ4ErVYLk8kEjUaDUqmEfD6PQqHAiYrIhQhDo9HAbreDMYZMJgMAkGUZjDFYrVa0trbiyJEjKBaLVdeh6yvPYzabYTAYUKlUEI/Hq64LoGoNAu8vCJJ6H2AuiQC5bJIkQa/Xo6GhAUuWLEF9fT10Oh0KhQISiQSGh4cxOjqKUCiEcrnMJ/mqVCo0NzfDYrFAkiR4vV7odDo4HA5EIhFUKhU4HA7Y7faqKbZEVnq9HgaDAR6PBy0tLaivr4fZbEaxWMTY2BgOHz6M4eFhFIvFKqIiCLJ6/0DEpOYRZzomRS4VkYwkSbBarVixYgVWrlwJxhgmJiYQi8Wg0+ng8/ngdruRzWaxZ88eDA0NIZfLoVQqgTGG888/HyaTCf39/eju7kZ/fz98Ph/K5TKGh4fhcrmQTCZx6NAhFAoFSJIEnU4Hm82GpqYmtLe3w263Q5IkDAwMIBAIQK1Wo6urC06nE0eOHMGuXbuQTCZ5rIviVUor60xCPC5nHsKSWtQgEqyWBxBZ2Ww2rF27Fo2NjYjFYjzYTUQwNjaGFStW4KyzzsKSJUswPj6O3t5ebuFotVrYbDbuNmo0GsiyjFwuh1QqhaVLl0KWZYTDYSSTSdhsNixZsgR+vx8OhwM2mw0TExMYHBzE6OgoCoUCGGPYt28fli5dijVr1sBsNuP1119HJBJBqVRCqVSqip0J0lj8ECS1aCFhejo3quQC9K/VasXHPvYxNDc3Y3R0FMPDwyiVSvD5fNBoNMhms0in0+jt7YUsyzj77LPR0tKC5uZmjIyMYHBwEBqNBvF4HIwxFAoFGAwGWCwWlEolJJNJHD58GM3NzfD7/ViyZAmam5vR1dUFq9WKSCSCPXv2YHh4GNlsFg0NDbBYLMjn80ilUhgZGYEsy2hsbESxWMRrr72GSCTCXVeVSnVMSYPA4oIgqUWLubVIZFGde+65WL16NQ4fPoy9e/ciHo+jXC6jVCpxosnlcigUCty66unpgU6n4+QxNTWF/fv3IxgMYunSpXA4HEin0xgaGoLRaESxWITBYMDq1auh0Wig0Wi4hbZv3z7s2rWLB8Oz2Sz0ej3UajXy+Tzy+TwymQzOOussNDU1IZPJcNevXC4DmM5IkusHCLJarBAktYihzN4pU/omkwkdHR081jQ+Pj4rk1YsFvlxyWQS+/bt47qofD7PY0/RaBT5fB7RaBRerxehUAiyLMPtdnOLzG63I5vN4rXXXoPNZoPT6cTQ0BDy+TwkSUKxWEQ2m+XXlGUZarUahUIBAwMDWLt2LZYsWYJ4PI4DBw4gn8+jVCotGJ2ZwHsLQVKLGMpAOWMMGo0GKpUKXV1dcDgcGBsbQzAY5OSgVquh1WqrXCo6PhKJYHh4GJlMBhqNBrFYDBMTE8hkMqhUKohGo2hubobD4cCKFSvw8ssvIxqNIpFIwGg0IhKJoL+/H263G1qtlruJSm2UWq0GME2SOp0OADA5OYlIJIK6ujrYbDY4HA5MTExwi5CIWMSoFi9Eq5ZFimN1J3A6nWhqakIul8PU1BSKxSJXgQNHH3IiNGCaQNLpNNLpNLLZLFQqFSYmJhAMBpHP58EYg16vh91uhyzLMJvNUKlUyOfzMBqNSKVSyOVycLlcMJlMKJfL3EWbKSug90iTRbops9kMrVYLt9vNtVxK7ZZSUiGwuCAsqUWImSJNCpbb7XY0NzdjcHAQhUIBwLSAc6YVorROKPVfKBQQi8VgsVggyzKSySTUajVMJhNUKhWcTie0Wi3UajXUajU8Hg9SqRRisRhyuRyampqg0+mQzWaRSqVQqVRQKpU42VC8iq5fqVR4cFyn0/HYk8ViQV1dHYLBIJLJZJUbK7ovLE4IklpkmFl7p9VqodFoYLFY0NLSwmNMo6OjMBqN8Pl8PL2v1WqriEPpTpGLZ7VaAYAHtskNs1gsyGQyyGazAMBdyEKhAK1Wi0wmw5Xo2WyWi0KVhcc6nY67fSqVCoVCAU6nk7t4iUQCPp8PpVIJBoMBExMTmJqaqopPCbdv8UGQ1CICPagkNZAkiQexm5ubuQtWLpeRy+WQyWTgcDhQX1+P0dFR7mYpCUalUkGn06GxsRGRSIRbUZQJlGUZTqcTXV1dnIjGxsYAAC6XC62trQiHwzAYDMjn83C73ZicnITL5eJBc8rWkcVEOi2dTgev14tcLodQKITBwUFUKhW4XC6k02k4HA4cPHgQgUBglmsrSGrxQJDUIsJMwaZGo+GB7FgsBkmSkMlk+D56vR6BQAB+v58Ht7PZLH/fYDDAbrfD7/fD6/XiyJEjMBqNGBgYQCwWg1arxcDAACqVCpYtWwaVSoX6+noEg0G43W6k02leDxgOh7m8QKVSwe/3o1wuI5FI8JYuRFLAdKyqqakJKpUKu3fv5jGtQCAAi8UCq9UKtVqNpUuXIpvNVgXi5+q0IFC7ECRVw5jZcE65Xa/Xw+l0YtWqVXA6ncjn8ygWi4jFYuju7kYikaiyStrb23nMiNTkFosF9fX1sNvtiEQi0Gg00Ov1yOVynAxIPkBCS41GwzsihMNhDA8Pw2q1olwuw2QyoVAooFwuw+1286C4zWbj3RPK5TLsdjtUKhU8Hg8ikQgymQyi0Si0Wi1SqRQXiZbLZVitVtTV1SGbzVYF5Mkamyaqo8JWgdqDIKlFgJntTqgJXWdnJ5qbm3H48GGoVCoEAgHEYjHs3bsXKpUKFosFPp8P+XwesiyjtbUV2WwWsiwDmLbGZFmGwWDggXiz2cwtIkmSeJEwkZjT6UQkEkE0GoXNZoNWq0UoFILb7ebFyESiS5cuRTweh8lkQnt7O+LxONLpNMxmM8rlMgqFAjKZDBKJBNd3lUolxGIx6PV66PV6SJIEu93Oi6KB2f2o/u9TgiCq2oQgqRoGeTOSVN0tU5IkWCwWXhw8NTUFjUaDcDjMH34qZSEiGR0dRVdXFycVsm4YY4hGozxDSNIECmxrNBoMDAygUCigq6sLlUoFfr8fwWAQ5XIZra2tmJqaQiqV4tk/rVYLnU6HUqkEh8MBo9EIk8nE3TSz2Qy73Y4dO3YgFovxYmY6b29vL2w2G+LxOCdkj8dTJUpVWpdHvT5BVLUIQVI1DXrgqrVBarWaB6qLxSIXQcZiMZ6Ny+VymJiYAGMM8XgcarUawWAQZrMZ4+PjCIVC8Pl8yOVyvB2wxWKBXq+vUoeTFebxeLgF09bWhuHhYYRCIf6+RqNBMBjk1hcF3gHAZrNhfHwc+XwearUara2tvNxGo9Ggp6cHDocDAOD3+3kmktai0WjQ1NSERCKBWCw2KyYlSUqiEqg1CJKqYdCDqJQJlMtl6HQ6WK1WuFwuLikgK8Tv9wMAkskkvF4vMpkMdDodlyF0dHSgq6sL0WgUqVQKqVQKHo+HW1WhUIhbQ8C06LO+vh4ajQb5fJ43wSO3MBwO87o8ki3IsozBwUGkUino9XpEo1Ekk0kYjUZ0d3djYmICe/bsQTQaRU9PD+/8CQBWq5WX5/j9fuzevRvJZBJNTU1wuVxIJBK8pg+AIi4lUKsQivMaxcyCYbJiNBoNfD4fXC4X73ZJbVFcLhfy+TyCwSAmJiag1Wpht9u56jwej2NgYAA2mw3t7e2wWq3I5XIwm81oaWnhrVXK5TIYY1Cr1WhpaYHZbOZxKCKktrY2WCwWLjUoFArQ6/XQ6XTI5XLo7e3F5OQkstksotEoSqUSGhsbUSqVeHeEhoYGtLa2wmg0QqvV8l5X5NY6HA44HA4kk0kkEglYrVbodLqqkpk5Prkz+4cSeNcQJLWIQCptn8/Hm8kxxnjgW6vVYnJyEmNjY7DZbPD7/bDZbDCZTACmA+XpdBr79u2DJEmw2WzcVauvrwdjDMFgkLtT2WwWhUIBIyMjKBaLKJVKyGQyCIVCyGQySKVSkGUZxWIRkUgE8XgcIyMjYIzB7/fDZDLBarUilUrBZrMhn8/jzTff5G1a3G43ZFlGoVCA1Wrl6ye1u0qlgtvt5iUzdB+i1fDigiCpRQJ6aKksxWQyIZVKIZ/P8yA1ZcqmpqYgyzKXGVCsiYisUCjg8OHDAICmpiYwxjA5OYlwOIxKpcKvRdoqKq0pl8s4ePAghoeHMTU1heHhYQQCAQBHM4VkVZFCXaVSoVwuo1wuY8+ePejr64NKpUJdXR1v3VIqlVAoFCDLMu9zlU6neYtit9sNn8/H6wZJzV49VGL6R6D2IGJSiwDKWjeLxQIAyGaz0Gq1vEaONE3pdBp6vR4Oh4NrpZQunFK1nkwmebbvwIEDaGxsREdHB/r6+pDL5eB0OtHa2opQKIRYLIZMJoNAIACtVguXy8WlAVTO0tDQAI/Hg3A4jHQ6DYPBwFXte/bs4QXJFFMqFotcc1Uul2E2m5FOp3nPdL/fD71ezwumZVmG1+vl2ci5IayrWoMgqRqFUn5AIEvI4XBAlmVeBpNOp7maO5fLob6+nsebpqamuICSSlRICU7Zs1KpBJ1OB71ej8nJSQBAQ0MDdDodjEYjstksRkZGAEzX7BUKBe6SeTwemM1mpFIpblWRRMJqtSIcDiMSicBgMMDr9fJ1ZzIZhMNhxGIxbiEZDAbeayqbzSKTycBoNMJqtSIej/POoKSZUlpSjB2Vagg3sLYgSKpmcVR+oBysQJ0zST0OgBf+JhIJaLVarikqFovcKiF3T5IkLg9IJpNwOp28TXAulwNjjBclj4+PY3h4GIlEokqFTpm1crmMsbExTnAajQYTExPwer180rEkSXA6nZBlmdfwFYtF7rKWy2XE43FOvhaLhWun6H6oYwPFymw2G9Lp9AxCIuW5QK1BkNQigFK8aLVa+Rw8ilFJksT1Qw6HA6lUCtFoFAAQDofhdrt5KYwkSbz7ZrFYxMTEBO+SOTIygqamJhw4cIC3YCG9FLV1UbZOIdIjoqD42NTUFIxGI5qamlAqlbjbqdVqqzqCUuzLbDZjcnISxWIRXq8XZrMZ8XgcwWAQFouFl+vQ+e12OyYnJznpKT8jgdqDCJwvAlC63WAwwGAwoFQqIZvNcklCoVBAOBzmcSSKH5XLZfh8Pp4dozHpNH6K4jqFQoFPJybRJj3wdByAKmtGWeisjIsB4JqmQCDAC4yVxKZWq7lcgWJlAHi7Fo/HA7VajXQ6za07svhIzT67LEagViFIqkYxlxaIWu5SoS25btRaxWw284yZ2+3mJSkqlYqTE1lkZMVQUF6r1UKr1Va1+QWqxZJENMpe6UpRZV1dHXp6erjLpuzAqSy3oftTWmKkmg+FQpAkCW63GyqVigfJKXNI1ya91MwSGYHagyCpmkR1xwNlzZ4kSTw+pNFoeMq/XC5Dr9ejUqlgxYoVsNvt0Gg00Ol0vNmdsiOnUiRKPaMaGxsxPj7Oa/CqA9OsyqJSvqdWq+H3+3HWWWfB4/Hwtio0zl1ZIqMkRoqxkbXmcrk4wdntdhiNRoTDYUxOTkKn00GtVnOy0ul0VfchULsQJFWTYHNaL2Q9JJNJ3ne8VCohEonwVH+xWOS1eHR8qVTiD7hOp+MPvDJTSG1/ZVlGT08PL4UhC0tpXdHaSBvldruxadMmLF++HKVSCfF4nKvclT3NlTov0k+RdUcBdKvVikQigXQ6zYmW7q9cLqNYLEKv1/N+U0qSEmRVmxCB80UC6gZAZEMPNhGWx+OBSqVCLBbjBAEcjSNR0a8kSfD7/TAYDFxiAEzX+g0PDyMej0OWZaxcuRKSJGFwcBDA0Wkv5OIpB5KuWrUKF110ES8Gpn2VE2yUKnGlVUXBb5pcHI1GUSwWMTY2htbWVhgMBgDA1NQUv7Zer4fNZsPo6OgZ/isIvBcQJLVIQFYLxY90Oh1vFqfX6+H3+zE6Olrl1hEJUJyKMYYlS5bA5/MhlUrx+jzGGBobG+Hz+TA4OIhgMAi1Wo3ly5ejUChgdHSUt3ahrCBZavX19bjkkktgs9kQi8WQz+cBgFteZLUROVGgnMiGLC0SppLyPJlMcvFnIpHgAlGdTsf1YsJyWhwQJFWTmO6LNDNATeUpBoMBjDGMj4/DZDLx7pnhcLjKegHAra5sNovOzk40NjbCZrPBbrdjamoKsVgMAHi20OPx8A4IBoMBmzZtwttvv811ScFgEJlMhpPKihUr0NXVhUKhAJPJxK0tCnaTi0cgMSltNxqNyOVy3PKjmJler0coFEJjYyPy+TwqlQqSySQcDgeXUZCrJ9oJ1zZETGqRgCbDGAwGaDQaJJNJZDIZFItF6HQ6xGIxxONxPu6ctEWU2bPb7XC73dDr9TCbzVxTRVIAk8nEawC9Xi9cLhcymQwuvPBCfPKTn4ROp+ODPEmMqdFo4HK5oNVqYTabUSgUEAqFePxIKR4lC4rWRtYVxacAVI3pIk1VoVCAy+UCYwwOhwPlcpnrpcj9pesI1CYESdUwlFokSZKQTqd5XIfa+eZyOdhsNl7mQoM1SVNFx7pcLhiNRthsNt6bPBKJ8P5Q9B5lCI1GI4rFInbt2sXLayKRSNUkGgCwWCxobm6Gx+NBIpFANptFLpdDqVRCLpfjI9NJS6XVanlrFmVcSxkAp6ykRqNBNBqtGsFFpT2kvJ+ZgRSoPQh3r8ahlB4o3Rqqh6P2JZlMBi6XqypzRjIFGnrQ0tICo9GIeDyOZDLJXSwqWiYtFUkb/H4/9u3bh0AgAJfLhXK5zJvoWSwW2O12NDU1YWpqCtFolBcyk7VERETWnDLTSMQ0c2AoWVi5XA5qtRqpVArxeJy7exaLhWf5NBrNrM9KEFXtQZDUIgBZTxaLhc/NIwvIYrHwjghGo5HHr0qlEi8cptgQteStr68HABw6dAjJZBIejwcOhwPFYpG7Z0RS2WwWjDGce+65OHjwICYmJiDLMurr69HT0wOXy4VwOIx8Ps+Hgiqnw1QqlSprCgDXOQHgsSlllwZyF4mEIpEImpubeX8rlUoFvV7PpRF0z4KkahOCpGoSyj5J1R06afin0+nklsXY2BgymQxMJhP0ej3XJFHmjPqd2+12mEwm5PN53s2TymnooacMIpGLx+NBJpPhAxh0Oh1GRkbw+uuv480334TP50NHRwfa29thNpvhcrl4b3WKMVEgnqQRRJ5ERKR0p32U65ckCfF4HMB065ZEIoHm5maMjY3B5XJhcnKSZzRFqUxtQpBUDUPp6lGNntPpRDQaRTQa5UHrQCDA3TCv11ulZSJBZSgUgt1uh9Vq5YFzGlUFgBckK7NxVHhMk2caGhqwZMkSdHZ24siRI+jv70dLSwvGx8exa9cumM1mOBwO3kKYtE90Xr1eDwA8rkVZR7LwlGU0ZAnq9Xqk02mkUilYLBaoVCpMTk4inU7DZrNBp9Nx8hOWVG1CkFQN4+g0FInHapQj0qnsJBqNciIxGo0866d0o0KhEO/Q6XQ6OeGtWrWKd/Q0mUxQqVS8mR0APl8vFApVTSxub2+Hy+XCpk2bsGPHDuzatYsTIo2vokA5/U7xKSWRkDZKq9Vyq4oymfQZUFth6gBBY9wpuyeGMdQ2BEnVOOjho4wXBaBJiT0xMcED3uFwGC0tLTyuRKRGGTGqgbPb7TzzRqUnuVyOk5hyKg2l/wcGBrg143K5eJD7xRdf5MMdSFJArmM4HEaxWKyy6Ig4yZUjgsnlcjwbSTE4tVrNy2Di8Ti/J5InkGKeRsyTFSYIq7YgSKpGMbMmTa1W8zgODUWQZRm9vb1VdXtkMZHym9TdJH4cHx9HqVTi7Vvsdjt8Ph+am5uRSqUQiUSq5t1RQF6WZYRCIQQCAdTV1cHlcuHAgQPQ6XRobm7m5SyVSgWyLMNutyOfz3MhJhEhWVMUpCdSIjIji4ruiTKFJFil9sOUDCCLT6B2IUiq5iDNGihAQxHy+TxsNhtUKhXvNKBso0uxJ1J/E0EpiaBSqWBkZITHfZqamqBWq7l8IZPJIJlMAgDXS6XTaciyjLa2NlitVrS0tGB0dBSZTIZbQ0r3kqa/qFQqTExMcPdzZluV6gGfR89DZEXWFnU8oI4N+XweU1NTqKur4+JWKtMRllTtQZBUjWJmYzmHw4H29nYYDAY+QEGj0SCdTle1P6HaOsoCAkcb1xUKBe5eUXD8L3/5C5LJJC/m9Xg8CIVCGB0dRTqdRl1dHVKpFDKZDHp6emC321EoFNDf38/1VGTh0O+kp/J4PEgmkzz7piyJ0Wg03AVUto0hq4u2Ub2fWq3mMwJ1Oh2SySTcbjfPaP7lL3/h1uP0uQAxlKE2IEiqhkEPqizL6OzshCRJGBoaQltbG+94QB04SbRJDyrFoihgTaptKn+pr69HMplEb28vGhoakMlkYLfbYbFYeM+mUCgEl8sFjUbDg+uZTAZ6vZ63MqYJNeQekkhT2SWUiIskAkRCSukAkRVZfTP7WZEUwev1cqmEXq+Hy+WCzWZDMpnEkSNHAOD/ziVGr9cKBEnVHI4OFFCWiZRKJezfvx82m413QYjH47ztCWXSyuUyH9BAww+SySTMZjNKpRJcLhd8Ph/a29sRCAS4XGBycpK3g9Hr9ejo6IDX64XJZILP50NXVxcf8ElB/HK5zOfjKTVRjDHE43EkEgm43W4Eg0FOSJTtA6Y7JFDMijKAZGEpWwsry4LI9SX3Lh6PI5PJoLm5GQMDA5zshMtXOxC1ezUMetASiQR27NgBvV6PZcuWIRaL8SD3zC6ZFFBOpVIIh8OIRqPIZDK8z5TFYsHq1athMpkwNTUFxhgCgQDGxsZ473NSjXu9Xmg0Gj7WKplMwmQyIRAIIJVK8WA2dVAgsSmpywcHB3l7FSIccukIVDpDxEuWmLKlCzA95IHGaDHGeJsaxhiSySRkWea9sUS3ztqCsKRqFMrBBWRZLF++nHciKJfLPPWuDC6T4pxiUMoWwy0tLVi5ciV0Oh12796NQCCAxsZGJJNJjIyMoLW1lcecQqEQ70JAVowsy4hGo5iamuISBgp0k/WitGQikQgSiQQnFiUJKvuqU5xMqTwvFArI5XLQarXw+XwYHh7mPaZKpRLy+TzPRFIw3WAwIJFIzCJugYUNYUnVKJTWgF6vR3t7OzKZDCKRCBobG5HL5fioKACc0AwGA5cXAOAZM5/Ph3PPPRd2ux1HjhzB+Pg48vk8CoUCPB4PJElCOBxGKpXiREOxJIqLpVIp9Pb2YnR0tKrBnVLbRMXEJJWIRCLI5XJcOkDxJnIZHQ4HD35T5wayriqVCrq6utDe3s6D7slkku9HwX+Px4NsNstHyyt7wgssfAiSqmGQZUJZN4PBwHVNALiwEzhaakIWD7VDqVQqcLlcWLFiBfx+P4LBIA4dOoRUKlVVvmK1WrlbyBiD2WyG3+9Hd3c3H9meSqW4FUMxpJnDECi+RGJPyvopyYeC6MpzUfcGim1VKhWYTCZOyNSmJplMIp1Oc+V9Pp/nxGg2m2d1RhBY+BB/sRqFUn5AKm5gmrjefvttPjJK2TSOxJdU81ZXV8cHbsqyjJGREd4cT6fTwefzIZfLQaPRcBKiuFAgEOAN8EZHR5FKpWC1WnkzPYo7kfs11/QWZa91cstohDr1hiK9F5EZkZdarYbBYMDIyAgmJydhNpt5FwSSUhQKBS4YJdGn8rMTqA0IkqpRKLNaVGZCGqOpqSkuLSANFBFUd3c3tFotRkZGIMsygOlC4UgkwvsykRRBOaSBJAFTU1M8ppTJZBAKhbjaPZPJ8HHsSv0STaAhfRRJIpRESxIIEn2Si0jnoKC7sgtCJpPByMgID/BTj3WlW2k2m2E0GmEymZDL5aoyeyIuVRsQ7l6NQqkWd7lcXENElga9pixYqVSCwWCA1+vFlVdeiZ6eHi5doInHNDZd6VLRQ03p/1wuh0AgUFXKQrV/ZG0VCoU5J9KQJaWcAkMaqHw+z69Fmi7an2JLyp5S9Jr6VJHbS+RptVp5+Q9NaiZtmCCn2oKwpGoQygeVsmI0LUaj0fCBB3q9nlsPFosF9fX1CIfDiMViaGpqQiAQ4FYMZfpoAgtZZcqMGrlb1D+KXD/KpCkLlmmNdCxZQ0Q4ygZ3VCqj7MapLC4mklSOyqL3aN9isch7YFG5jiRJ6O3tRTweRywWQ19fH78v0bqldiBIqsZRqVS4RkmpE7Lb7airq8P4+DjcbjdWrlwJs9mMeDyO4eFhtLe3w+l0IpvNclIhoqACX+AoCdJDTxoppWRAqWsiS4hIhUZjmc1mTkLKTJ/SUqPt9B5JEeg+KcamrO1TXpcsSFK0VyoV9PX14dChQ7wUSGnVCYKqDQh3r8ZRqVQQjUYRCoV4gJsKbf1+P+x2O8477zxceumlOPvss2Gz2Xiguauri5MaZdSIBJSdL5U1ckoRJckXqLwmn88jHo8jFAohFAphZGQE8Xi8qpe5EtTpAEDVtchlpX0oEE71iHTfZA1RcJ+uQcMmkskkcrkcUqkUJy3RW6r2ICypGoWyO0A8Hse+ffswPDwMAGhtbeX7mUwmdHV1oaWlhVtM4XAYarUa3d3diEajGBoa4uSQTCa5m8fY9Oh2ihsBqIoVEbkRwRSLRcRiMd5zymq1Vs3aU5KDkgyPJVeYS8uklFVQQoBaHptMJlitVh4voyC9MlAugua1B0FSiwBUC5dKpXhHSkq/Uw0bNcLz+XwYHR1Fb28vb8FCLVPGxsZ4dg0An9pCTe4AVMWFAHALhjpuUvdOyhySBUMupDKONBcJKRXn1FKYXD6qO1Sui/pTAeCFz9SjXdk2WEmSgqBqC8Ldq2HM1X+JsnhECKVSCRMTE9zSIX0RdfEcGRnhHSxJpkDnouOV1pIypgNUE5nSElIOACUQWcwMftN5KNOoPI9Go4EsyzCbzbxXFsWWaDqO2WzmsoZUKsXV5lQzONNCE6gtCJKqcSitA2oTrCzIZYyht7cXgUCAb2tvb+e9zNPpNM/WkUWkzOaRNaIc1KkMXpNYU0lmSumBcm0UEFcWEiuzfzqdjr+vHLvFGEMqleK1iKVSiceZEokELySmjpyUAMjlclXrmPm7QG1AkFSNQxnEJpV1Lpfjo9GtVisOHTqE3/3ud3w7tWORZRkOh4MfQwJIpTVDqnFlWxRloTDV4JE2igSXhUKBExdw1Ioi8lFKAJTBeKVCfiahKOsENRoNJicnqybh0GQYWZZ5GxrldQRB1SYESS0CKLNWNL6cHnSXy4VMJoNXXnkFr7zyCreWaEwUjb2iYLeSOIg0KEZE24h4lF0YKItGlpUyuE3EViqVuLSB3C9l1o+a4hFh0TUo00cWHBGSsqmf2+3mY9UdDgePxx2bmCRQXy6BhQ0ROF8EUGbbKAbV0dHBh3ra7XYEg0G88MIL8Pl8WL58OZLJJC9/oXbDVOhLUBIJWUcUKKeyG9qHsoHKeBNl2ZQWFQXlSX5ALp5Op+NDHZxOJywWCx/HRbIKOp56uE9NTfGeVUuWLAEAbgkq28jQZzTjU4MgqdqAIKlFAiVRjY2NYenSpVCr1UilUmhqakKxWEQwGMTbb7+NhoYG+P1+riMaHh5GKpXi+iISaiotKsrwUfkKkZLSnVOqwynGRGQFHA2OK4PZVPzLGIPNZkN3dze8Xi+MRiPeeOMNHoBnbLoXltFoxNjYGEZHR/laTCYTDAYDzyRqNBqMjo7yDqTC1attCJJaJFCm9FOpFEKhEJxOJ4rFIjweDxibHprZ19eHt956Cxs2bOA9yJ1OJ++kqXygybVTEhGV0CjjSNScThlrAsCzazQlBgAnMApy03EmkwlmsxkjIyM8czc0NIR0Os3dUKvVikQigYGBAd6axWKxoLu7m7uYlCwYHBw8CWW5IK5agCCpRYKZ5DI5OQmPx8MHK9TV1aFQKCCTyeDPf/4z/H4/PvjBD0Kr1fKJLTQ2nVw/ZaCbVOXKYt9SqYRkMgmNRsOHiJIcgdr5KoPYlB2k4HdbWxs6OzsRj8cxODhYFQQnDRfdj9PphNVqRW9vLyctGhjh9XoxNTWFfD7Ps3qhUGgWOQlrqjYhSGoRQWnFTE5OorW1lWfgli9fDlmWsXPnTrz99tt8jLrL5UI+n+c1dpTiV3Ys0Ol03N0j64TaECuze9SzieoJab4fkRz1vKJi51KphF27dqFUKiESiaBYLPLJxolEAhqNBk1NTbDZbDxLGQwGudvHGENdXR1isRgsFguSySS8Xi8mJye5q0cQBFW7ECS1SKCUBwBAMplEOByGzWZDNBrFwMAAn0OXy+Wwb98+PPXUU/j0pz8Nl8uFgYEB7s6RtUQ1c8rR6Er3jjJ51GnBYDAglUrxEVaMMR4MN5lMnEA7OjoAAIcOHUI8HgcwbZnRNVKpFICjXTWXLVsGWZarCoVdLhcaGhp4sTP1XDcYDBgcHOQZQfpsBGoXQoKwiKBUn5fLZUxMTMBsNvNe44lEAp2dnbDZbCiVSnj77bfxzDPP8N5LdBx1MlDOyUun01WBaZ1OB4vFgvb2dvj9fnR1deHCCy+E3+/nGUIKzJOEwWw245xzzoHdbkckEqnqe0UyBLLMaI5fMplENpvF/v37MTg4yAc8uN1u1NfXc9IMh8Pw+/1IpVIIBAJV7YoFahvCklpEUJIUAD61xe12IxqNwmq1wuFwYOXKlXj11VdRKBSwY8cO3toll8vxYDiJMZVtU5QWCWXWJEmC2+2GTqeDx+PBli1bEAgEsG3bNvT19fGiZZ/PB5fLBcYYDhw4gFQqBZVKxclR2eqX6g2pB/rY2Bj27dvHY2eNjY3w+/18MvKhQ4fgcDjgdruxbds2ZDKZKp2WIKrahsSELTxveC9qyYg4yAJqa2vD+vXrEQwGucVUKpWwe/duHDx4EKVSCbIs87bCM9upKJvTkYVFgW0AXGmu0WiwatUqnH/++dyKo2A4kQtZZ9lsdtZQBcoMAtXlNKlUCoODg8hkMlCpVDCbzTj//PNhMBhgs9m4fKKnpwfBYBC/+93vuKtJ5zidX3HxuJx5CHdvkUGZ4atUKhgbG0MwGERDQwPC4TAqlQri8Th6enqwZMkS7srt378fqVSqKvgOoIpMyG1Lp9O8TTBwdIjC5OQkgsEgzGYzWltb0d3dDZvNBuBoNnDmrL9MJsPb/1K/KIpN5XI5jIyMcNIxmUzo7Ozkx4fDYSSTSTQ0NHDiJdeWPgtBKrUPQVKLEPRwMsaQz+dx8OBB7rZNTk6ivr4eqVSKx5NoovHY2BgvqVGqzZXN6GZqp5SuYC6Xw9DQEFeK79q1C4FAgAfWCeRK0lolaXqYBF2rWCwiFApxCwqYzgh2dnbybg1GoxGBQAAulws2mw1vv/02xsfHBSktQgiSWoRQklSlUsHo6Cj6+vrQ2tqKbDaLVCoFm82GWCyG9vZ2WCwWSJKEaDSK0dFRHtMhAiLrhl6T4hwAlyi4XC4sW7YMZ599NkwmEw9okyWmjJfRa2U5DMkZSqUS4vE4xsbGkEwmUalUoNfr4fV6IUkSGhoa4PP5MDAwwOcN0qzAmWJUQViLAyImNY94L/sbKfsnaTQamM1mXHbZZbDb7ejt7YXBYIAsy0gkEohEIjh48CCfd2e329HQ0MBlA2q1mssPSEelbOPb2dmJ1atXw+fzQa/X8/5VarUaQ0NDeP311xGNRrl0QUl2VIJDBJPJZDA4OIh4PM5JrKWlBY2NjbDb7WhqasJbb72FQqGAtrY25HI5bN++HRMTE7O6f74XX23xuJx5CEtqkWKmNZVMJvHKK6+gVCqhvb2dWyl2ux2tra1Ys2YNr92LRqM4fPgwRkZGeGcDmg5DP0pCcDgcfDbf5OQknn76aTz//PNIp9NQq9VIJBI87kRroyk1pG4HgHQ6jYmJCSSTSQDTsS6PxwOLxQKn0wmz2YydO3eiWCyipaUF5XIZO3bswMTExCzrSZDJ4oGQICxiKIuOJUnC+Pg4Xn75ZWzcuBFtbW3o6+uD0+mE0WhEc3MzyuUyDh06hFgshmKxiPHxccRiMXR0dMBoNM7q90QBbQq4T01N4Y033kBvby9yuRyCwSC0Wi0CgQAPoFPgnMpkVCoVstkswuEwwuEwH9UuyzLq6+vhcrnQ2NgItVqN3t5eSJKE1tZWVCoVbNu2rSoOpbxfgcUD4e7NI85UO9uZPZtIeAkAIyMjkCQJPp8PqVQKQ0NDOHz4MA9YU3eCrq4uGI1Gvo1El3R+u92OcrmM0dFRPuKd2sE4nU60trbyALuy1XAkEkEwGEQ0GuVdQC0WCxoaGvhPNpvFxMQEbDYbb9L3pz/9CcPDw1x8eqYsKPG4nHkIkppHnCmSUsZ/APAWwmvWrIHD4cD4+DhSqRQ8Hg9yuRwOHz6MYDDI6++oZKWpqQkul4sH04FpacHk5CTS6TRvSUylKyQxKBQKaGxshMFg4EHzdDqNZDKJqakp3r5YrVbD6XSisbGRE1Q0GsWRI0fQ0NCAuro6ZDIZ/OEPf8DY2FiVSwsA0/8IklpsECQ1jziTgwGUrVwocG00GrFq1Sr09PSgWCxiaGgIRqMRarUa0WiU655IjKnVauH1enkMiromUP2d3W7nAfZsNssV60R2yoZ41KOc9tFoNPB4PGhqakJbWxscDgdGR0cRiUTg8XjgdDoxPj6ON998E5OTk7MC5GeCoOhaAmcWgqTmEWd6eomSoCi2pFKpUF9fj3POOQdarRZjY2OwWCzw+XwYHx/H+Pg4QqEQwuFw1ZBOrVbLz6FWq2E2mwGgSqKgtHJoeGgmk+GFybRdo9HA6XSiu7sbLS0t0Ov16OvrQzKZRFtbG8rlMvbu3Yvh4WHeowqorss7U19j8biceQiSmkfMF0kp9UvUCthkMmHdunVwuVzo7++Hy+WC2+1GOp1GOBzmJS65XG5Wlo7IikhDKQCd2TNdWfpCGT6TyYRly5ahtbUVU1NTmJycRKlUQl1dHSYmJjAwMMCzg0oR6HwoysXjcuYhSGoeMR8kRVaOVquFLMtQq9UoFovI5XIwmUxYu3Yt6uvrebsTn8/Hu3GGQiHEYjFMTExwVw04OrFGOfKKQESiJCkiJqvVisbGRphMJpTLZQSDQaRSKe5yjo+PI5lMchX6zHFZ81E4LB6XMw9BUvOI+RhWSdaUXq+HwWDgE4KLxSIfHnr++eejoaEBqVSKt0qhti0kRYjH4ygUCrxZnlarRblc5sNHAfA6OirPoWGhRqMRsizzIabRaBS5XI4PbYjH4zzbpyxCntkOeD6+uuJxOfMQJDWPmK+JuuTmUSGvMqBNAe6WlhYsW7aMt/CNRqOIx+Nc3wRMiy1LpRKsVisfhKDX63lZDMWeaDx6IpFAPB6fVQRNdX+RSIRn+uhHmZWc2SVhPiAelzMPQVLziPkiKWU7F2V9nnKYZqVSgSzLcLvd8Pl8MJvNMJlMkGWZtwNOJBK8eJgkCfl8npfXUPYPOCq0JPcyn8/zRnqFQoFPp5mdtTv26/nAfF///QhBUvOI+SIp5fWJYKhgGADP3FEbFxqzrlarYbPZ4HK5uLtoNBq5JQZMd/8ka0gZ3C4Wi0gmk3wqjXJIKFBdzjJb/7RwSl0WwhrebxAkNY+Yb5KiNVCcigZ/Kq0rilNRvV42m0UulwOAqv2VLiNl/4DqADd91WaWrsz8CiprDo+1z3xhoazj/QRRu/c+Bz10VF6ilBMoyYWC4wCqsm1KQlLO15t5flKaK+NMSrIiglMOGVUeL/D+hbCk5hELwZIi0FqU7h/V2AGosrAYY1Wz+Gbqo44ltiRiouvQuC16vRAtp5lYqOtazBCWlACA6g4CVKqitJ4A8MA6CULJHSSrai4XbmaHAuCou6cM0h89DhCThQWUEJbUPGIhWVJKKNXiJPykvuLKWBG5cEReJE+goPmxXD/6fa6g+UJHLaxxsUGQ1DxioZIUUO3+Kev9KOunJBilJUQuHLmBc8kGjhc0X+iotfUuBgh3T2BOKB9GspRIAT5zQANw1Kqaa87d8UhKQOBEEJbUPGIhW1JzQblepaU18z3lV2ohqMRPJxbLfdQShCUl8I4wMxA+M2t3rP0FBE4VgqTmEeIBFhA4McS0GAEBgQUNQVICAgILGoKkBAQEFjQESQkICCxoCJISEBBY0BAkJSAgsKAhSEpAQGBBQ5CUgIDAgoYgKQEBgQWN/w+qugG37iNBVAAAAABJRU5ErkJggg==\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
],
"image/png": "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\n"
},
"metadata": {}
}
],
"source": [
"visualize_model(model_conv)\n",
"\n",
"plt.ioff()\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "ZbmJT9frW2lF"
},
"source": [
"## Further Learning\n",
"\n",
"If you would like to learn more about the applications of transfer learning,\n",
"checkout our [Quantized Transfer Learning for Computer Vision Tutorial](https://pytorch.org/tutorials/intermediate/quantized_transfer_learning_tutorial.html).\n",
"\n",
"\n",
"\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.10"
},
"colab": {
"provenance": [],
"gpuType": "V100"
},
"accelerator": "GPU",
"gpuClass": "standard"
},
"nbformat": 4,
"nbformat_minor": 0
}